DayVectors

* 23 entries including: Joe Biden (series); generative AI arrives (series); human brain map; Putin sabotage campaign against NATO; Hunter Biden court cases | Trump in court | Biden works with big business; honeycomb wind turbines | Osmo Pocket 3 camera | climbing cranes; European farmers & climate change; Ukraine War disinformation | MAGA criminals | solutions for Gaza?; synthetic DNA crackdown; PAC-MAN clone | Tories failing | Spoutible rising | Trump's tax problem; Ukraine drones; Russian economy can't sustain Ukraine War; Trump hush-money trial | new munitions to Ukraine | Macron speaks; lower threshold of stardom | dangerous proteins | PACE in orbit.

[FRI 31 MAY 24] JOE BIDEN (28)

* JOE BIDEN (28): The resolution of the SCOTUS nomination issue was something of a relief to Joe, but he was still deflated from the humiliating failure of his presidential bid. The fiasco had been hard on Jill as well, with Delawaran expressions of sympathy for Joe making it troublesome to even go to the supermarket. The couple would commiserate some evenings, with Jill telling him one night: "It's hard to smile."

Joe replied: "I know. Things will get better."

He did get a sympathetic review from David Broder of THE WASHINGTON POST, who had asked to talk to Valerie off-the-record; she was reluctant, Broder having been critical of Joe in the past, but finally agreed. Broder's piece in WAPO declared: "The [media] focus has been too narrow to do justice to the man." Broder went on to say that though Joe was "impulsive" but "not all of his instincts are self-serving", going on to describe his compassion and empathy. Indeed, people who knew him well often said he wore his heart on his sleeve, and he was totally sincere.

Following that, during Christmastime 1987, Joe sat down for a long interview with Broder, who returned the favor on New Year's Day with a flattering, kindly, and perceptive opinion piece:

QUOTE:

END_QUOTE

Joe admitted to Broder that he had been "cocky", "immature", and "naive" in his presidential race; in response to failure, he had doubled down on his Senate work, doing an impressive job on the Bork hearings. Broder commented that his efforts on the Foreign Relations Committee, in the face of the fall of communism in Europe then in progress, promised to be just as significant. Broder concluded that ...

QUOTE:

END_QUOTE

As Ted Kennedy had said, there was life after a presidential campaign. Joe went on with his work with his usual determination, going on a tour of Europe in January to speak with leaders about the INF Treaty. After he came back, something went wrong. While lifting weights in the Senate gym, a severe pain shot up through his neck and into his head. He took the train back to Wilmington that evening, to get another shot of the pain, leaving his right side numb. He was concerned, but the doctors just concluded he had a pinched nerve, gave him a neck brace, and suggested exercises. It didn't seem like anything serious. [TO BE CONTINUED]

[THU 30 MAY 24] HUMAN BRAIN MAP

* HUMAN BRAIN MAP: As discussed in an article from NATURE.com ("This Is The Largest Map Of The Human Brain Ever Made" by Gemma Conroy 12 October 2023), researchers have created the largest atlas of human brain cells to date, featuring more than 3,000 cell types, many of which were previously not known.

The expansive cell atlas provides unprecedented insight into the brain. Researchers have previously mapped the human brain using techniques such as magnetic resonance imaging, but this is the first atlas of the whole human brain at the single-cell level. The research is part of the US National Institutes of Health's "Brain Research through Advancing Innovative Neurotechnologies Initiative / Cell Census Network (BICCN)", a collaboration between hundreds of scientists. The program's goals include cataloging brain cell types across humans, non-human primates, and mice to improve understanding of the cellular mechanisms behind poorly understood brain disorders.

Kimberly Siletti, a neuroscientist now at the University Medical Center Utrecht in the Netherlands, and her research team laid the foundation for the atlas by sequencing the RNA of more than 3 million cells from 106 locations covering the entire human brain, using tissue samples from three deceased male donors. They also included one motor cortex dissection from a female donor that had been used in previous studies. Their analysis documented 461 broad categories of brain cell that included more than 3,000 subtypes.

Neurons varied widely in different parts of the brain, suggesting different functions and developmental histories. The mix of neurons and other cell types also differed across each region, with some cells only found in specific locations. The brainstem -- which connects the brain to the spinal cord, and which has been under-researched -- turned out to feature a particularly high number of neuron types.

Related studies inspected the mechanisms of gene regulation and expression in different cells. Joseph Ecker, a molecular biologist at the Salk Institute for Biological Studies in La Jolla, California, and his colleagues investigated the brain tissue samples from the same three donors. They analyzed chemical markers that switch genes on or off in more than 500,000 individual cells. The switch molecules allowed the team to identify nearly 200 brain cell types. Even the same gene in the same type of cell could work differently across the brain. One gene was turned on with one switch at the front of the brain and with another at the back.

Pinpointing the switches that activate or block gene expression in brain cells could prove useful for diagnosing brain disorders and developing tailored treatments, according to Ecker: "That's another tool that comes out of the toolbox we're building."

Bing Ren -- a molecular biologist at the University of California, San Diego -- and his team similarly analyzed how more than a million brain cells from the three donors access and use genetic information. The investigation uncovered links between certain brain cell types and neuropsychiatric disorders, including bipolar disorder, depression and schizophrenia. The researchers used the data to predict how the genetic switches influence gene regulation and increase the risk of neurological diseases. For instance, in cells known as "microglia", which clear away dead or damaged cells, the presence of some genetic switches was strongly linked to risks of Alzheimer's disease.

Ren says the next step for the BICCN team is to sequence more cells from all parts of the brain. The researchers will also work with more tissue samples to build a picture of how the human brain can vary across populations and age groups. Ren says: "This is only the beginning."

[WED 29 MAY 24] GEN-AI ARRIVES (16)

* GEN-AI ARRIVES (16): Despite the challenges with using GAI chatbots in education, many researchers, educators, and companies think it has enormous potential in that field, and are experimenting to find the best use of the technology. Some use alternatives to ChatGPT, which is the current standard; some find ways to minimize inaccuracies and hallucinations; and some improve the chatbots' subject-specific knowledge. Collin Lynch -- a computer scientist at North Carolina State University in Raleigh, who specializes in educational systems -- says:

QUOTE:

END_QUOTE

Sobhi Tawil -- director of the future of learning and innovation at UNESCO, the United Nations' agency for education, in Paris -- suggests that society needs to help students to understand the strengths and risks and chatbots, and not just forbid their use. UNESCO has recently published a report titled "Guidance for Generative AI in Education & Research", one of its key recommendations being that educational institutions need to validate tools such as ChatGPT before using them in learning environments.

Companies are already marketing commercial assistants, such as MagicSchool and Eduaide, based on OpenAI's LLM technology, to help schoolteachers plan lesson activities and to assess students' work. Academics have produced other tools, such as PyrEval, created by computer scientist Rebecca Passonneau's team at Pennsylvania State University in State College, to read essays and extract the key ideas.

With help from educational psychologist Sadhana Puntambekar at the University of Wisconsin at Madison, PyrEval has scored physics essays written during science classes by about 2,000 middle-school students a year for the past three years. PyrEval provides individualized feedback to the students for the essays, and also provides feedback to teachers to allow them to tweak their class offerings.

The AI tutor and teaching assistant Khanmigo is more ambitious. It was generated by a partnership between OpenAI and education non-profit organization Khan Academy in Mountain View, California. Using GPT-4, Khanmigo provides students tips as they work through an exercise. It appears as a pop-up chatbot on a student's computer display, allowing students can discuss the problem that they are working on with it.

As the tool passes on queries to GPT-4, it adds a prompt to tell the bot to not provide answers, instead to ask lots of questions. Kristen DiCerbo -- Khan Academy's chief learning officer, calls this process a "productive struggle" -- but admits that Khanmigo is still in a pilot phase, and there is a fine line between asking constructive questions, and simply frustrating students.

Khanmigo was first introduced in the spring of 2023, with tens of thousands of students using it during the school year. Users included private subscribers as well as more than 30 school districts. Individuals paid $99 USD a year to cover the computing costs of LLMs, and school districts paid $60 a year per student for access. To protect student privacy, OpenAI agreed not to use Khanmigo data for training.

Khanmigo is a work in progress, however. LLMs are, again, prone to hallucinations; the prompt that Khanmigo sends to GPT-4 was revised to include the right answer, to keep GPT-4 on track. When it still failed, users were asked to tell Khan Academy about it. [TO BE CONTINUED]

[TUE 28 MAY 24] PUTIN'S SABOTAGE CAMPAIGN

* PUTIN'S SABOTAGE CAMPAIGN: As reported in an article from NBCNEWS.com "Russia's Brazen And Intensifying Sabotage Campaign Across Europe" by Dan De Luce and Jean-Nicholas Fievet, 13 May 2024), it comes as no real surprise that Russian President Vladimir Putin, confronted with NATO support of Ukraine in its struggle against the Russian invasion, is fighting back.

Sabotage attempts have included an alleged Russian-backed arson attack on a Ukrainian-linked warehouse in the United Kingdom, a plot to bomb or set fire to military bases in Germany, attempts to hack and disrupt Europe's railway signal network, and the jamming of GPS systems used by civil aviation. Physical attacks are coupled to cyberspace attacks, including hacking and the long-standing flood of Russian propaganda and disinformation.

Oleksandr Danylyuk -- of the Royal United Services Institute, a British defense and security think tank -- says: "It's very disturbing, and it's not like Russia has finished this process. It's still ongoing. Russia is definitely at war with the West."

German officials recently announced that they had uncovered an elaborate hacking campaign by Russian military intelligence that compromised the email accounts of the country's Social Democratic Party headquarters. Russian hackers also targeted German companies in the defense and aerospace industries.

Outside of Germany, Russia has staged thousands of cyberattacks on Czech and European railways, including hacking into signals and ticketing services, according to a Czech official. So far, there is no indication Russia has managed to seriously disrupt the supply of weapons, ammunition, or other aid to Ukraine -- but a Biden Administration official said the attacks were becoming more aggressive and that Russia was "crossing new lines".

Russia has denied everything, with Kremlin mouthpiece Dmitry Peskov saying: "All these statements, all those demarches on the part of European capitals are totally baseless and we decisively refute all of them."

Danylyuk, who has advised Ukraine's government, was one of the authors of a report issued in early 2024 that warned of a growing threat from Russia's military intelligence service, the GRU. The report said the GRU was setting up a covert network of operatives to conduct espionage and potential sabotage operations in Europe: "The GRU is restructuring how it manages the recruitment and training of special forces troops, and is rebuilding the support apparatus to be able to infiltrate them into European countries."

James Cleverly, the British home secretary, reported to the House of Commons that Russian activity ranged from plans to damage military aid bound for Ukraine in Germany and Poland, conduct espionage in Bulgaria and Italy, and mount disinformation campaigns to influence the outcome of European Union elections in June. Cleverly said: "Over a number of years, we have witnessed Russia and its intelligence services engage in yet more open and more brazen attempts to undermine our security, harm our people and interfere in our democracies."

Cleverly also said that a Russian diplomat in London had been expelled after being identified as a GRU officer. British prosecutors have accused a British man, Dylan Earl, 20, of masterminding an arson plot on a Ukraine-linked commercial property in March after allegedly being recruited as a Russian spy. Four other suspects have been charged in connection with the case.

In Germany, authorities have accused two German-Russian nationals of planning to target military and logistical sites, including US military bases, on behalf of Russia's intelligence services. The head of Germany's domestic intelligence service, Thomas Haldenwang, has warned that Russians have scaled up their ambitions: "We ... assess the risk of state-controlled acts of sabotage to be significantly increased -- especially against critical infrastructure in information and communication technology, as well as in energy supply,"

A Biden Administration official said: "We are prioritizing efforts along with our allies to track these Russian activities, disrupt and expose them." -- adding that "there have been significant successful European law enforcement efforts in recent months."

As Danylyuk points out, none of the sabotage has come close to doing as much damage to the Ukrainian war effort as the recent six-month delay in US military aid to Kyiv, engineered by Putin sympathizers in the US Congress. European officials worry that Russian political-influence campaigns are being conducted in advance of European Union parliamentary elections in June to boost far-Right parties that oppose arming Ukraine.

In April Belgium's prime minister, Alexander De Croo, accused Russia of trying to bribe members of the European Parliament to support its efforts to reduce European support for Ukraine. De Croo said a Belgian investigation "shows that Moscow has approached and also paid European members of parliament in order to promote the Russian agenda."

[MON 27 MAY 24] THE WEEK THAT WAS 21

DAYLOG MON 20 MAY 24: I've just finished reading Hunter Biden's memoirs, an appalling narrative of a three-year binge of booze and cocaine that's not such a pleasant read. He wrote it mostly to publicly confess and set the record straight.

HB is clean & sober now, grappling with two Federal court cases -- one for illegal possession of a firearm, the other for tax delinquency. HB had come to a plea bargain that let him off easy, but the courts didn't like it, so the Feds threw the book at him.

I got to wondering what to make of the court cases and dug into legal analyses online. The firearms case was based on the fact that he lied when he bought a pistol, saying he wasn't a drug user. Incidentally, Hallie Biden -- the widow of Hunter's deceased brother Beau, Hunter and Hallie having got together after Beau's death -- found the pistol and got rid of it. Given how unstable Hunter's behavior was at the time, Hallie must've found the pistol frightening. In any case, while Hunter is facing a maximum penalty of 25 years, the reality is that nobody ever does time in such a case if they don't have a rap sheet.

It's pretty much the same for the tax case. The IRS typically just wants their money, with penalties applied, and HB made restitution. It is very unusual for such a case to continue after that. One suspects the defense in both cases will be on the basis of "selective prosecution".

Incidentally, HB managed to get out of his doom spiral in 2019, when he met Melissa Cohen, an activist and film-maker of South African origins. They got married only a week after meeting. At the outset, he told her he was a crackhead; she replied: "Not any more."

Once they got together, MC went through his bags and threw out drug kit; when HB's pusher "friends" showed up at the door, she told them to get lost or she'd call the cops. She wouldn't let him go to the bathroom at night unattended, she changed the password on his notebook so he could only get in if she let him, and when she had to leave him alone, she'd take his wallet, car keys, and phone. It appears she was experienced in dealing with addicts. Aside from the court cases, everything seems OK now.

DAYLOG TUE 21 MAY 24: Donald Trump's Manhattan "hush money" trial has gone to the jury, with a verdict likely next week. On Monday afternoon, the testimony for the defense went off the rails.

Attorney Robert Costello, a longtime Trump ally, testified about a meeting he had with Michael Cohen -- the central witness for the prosecution -- in April 2018 after the FBI raided Cohen's home and office, with Costello attacking Cohen's credibility. The prosecution kept raising objections to Costello's testimony, which Judge Juan Merchan sustained. Costello could be heard sighing theatrically and saying "ridiculous" or "jeez" from the witness stand.

Not even 15 minutes into his testimony, the judge called a stop and sent the jury out of the courtroom. The judge then read the riot act to Costello:

QUOTE:

END_QUOTE

Costello said he understood, but Judge Merchan did not like his demeanor, angrily asking: "Are you staring me down?!" The judge then told all 60 reporters to clear the courtroom. The court transcript had him saying to Costello and defense attorney Emil Bove:

QUOTE:

END_QUOTE

Bove said he understood. Costello asked if he could "say something," but the judge cut him off: "This is not a conversation." Costello replied: "OK." After the exchange, everyone was brought back in, and there were no further problems.

Legal observers noted that it was nothing unusual for defendants to get into fights with a judge, but Costello is an attorney and should know better. In any case, we have to wait on the verdict. I'm not betting it will go well for Trump.

Incidentally, Tony "The Mooch" Scaramucci, a prominent Trump critic from the Right, commented recently: "My wife hates him almost as much as Melania does!" Oh, the shade!

DAYLOG WED 22 MAY 24: The conversation on Spout got around to the currently Right-dominated Supreme Court, and what to do about it. There's been talk of "court-packing", but it's not a good idea -- each side would pack the court on taking office, reducing SCOTUS to chaos. An 18-year term limit would do the job. By the time it was implemented, Clarence Thomas, Samuel Alito, and John Roberts would be over the limit and gone. SCOTUS would then shift from 6:3 Right to 6:3 Left.

I commented that Roberts wasn't particularly a problem compared to Thomas and Alito, a claim that was challenged: I'zNt RoBerTz aZ bAd aZ the oTHerZ?!!! No, not really; he is generally respected as a person, and has judged toward the middle many times -- saving ObamaCare for example. The problem with Roberts is that he is a conservative, and conservatism is dying. If it was ever credible, it isn't now. Roberts needs to go into honorable retirement.

That got some flak as well, to which I replied: "I'm willing to give the benefit of doubt." Yeah, I'm intellectually cautious, have some skepticism of what I hear, and prefer to focus on what I'm more certain of. I said: "I'll even give TRUMP the benefit of the doubt." That definitely got some flak, to which I replied: "I don't believe Trump is guilty of ALL the things he's accused of. Just most of them."

DAYLOG THU 23 MAY 24: The fighting in Ukraine roars on, but Ukrainian resistance is stiffening. There's no shortage of shells now, and the AFU has been steadily knocking Orc Su-25 attack aircraft out of the sky -- the Su-25s, it seems, being used to drop KAB glide bombs. The Su-25s won't get near the front lines, so it appears the AFU is being stocked with long-range anti-air weapons -- exactly what kind is unclear. F-16s are supposed to arrive soon, which hints that they may already be in the fight, firing newer long-range AMRAAM missiles.

That is welcome to all except Putin, but Ukrainian President Volodymyr Zelenskyy is not happy, saying: "Our partners fear that Russia will lose this war." -- and that weapons deliveries are always a year too late.

What the Allies fear is that, in response to provocations, Putin will start launching missiles at the Baltic States, Poland, even Finland. Like Putin wouldn't do it? Yes he would. As far as arms deliveries go, NATO was not ready for this war and did not have so much to spare. Ramping up production can't be done overnight. A fundamental rule of warfare is that the time required to build up to attack an adversary gives the adversary the exact same amount of time to prepare for the attack. Getting ahead means finding an edge.

Of course, Zelenskyy's hot talk may be partly disinformation, with a different story being told behind closed doors. The hot talk may also be signaling a change in Allied doctrine. As an interesting possibility, could the Allies start conducting air patrols over Ukraine to shoot down Russian missiles? That sounds plausible. French President Macron may be the signpost for where things are going. If he starts talking up actions, they may already be in motion.

As a footnote, it appears that the Russian offensive that jumped off on 10 May was moved up in response to resumption of US arms shipments to Ukraine. That tends to confirm my suspicion that Putin is conducting a "go for broke" offensive -- knowing that time is not really on his side.

DAYLOG FRI 24 MAY 24: The 6JAN21 Capitol riot has led to charges for about 1,400 participants, with almost 900 being sentenced and two-thirds of them receiving jail sentences ranging from a few days to 22 years.

Today, sentence was passed on Christopher J. Quaglin -- an electrician from New Jersey, who was convicted of assaults on police officers during the riot -- by District Judge Trevor McFadden, who incidentally was appointed by Donald Trump. Quaglin had taken pride of his role in the riot and was not contrite with Judge McFadden, even attacking the judge: "You're Trump's worst mistake of 2016." The judge suggested that such remarks were a bad idea at sentencing, with Quaglin persisting: "It's a kangaroo court."

Judge McFadden gave at least as good as he got, calling Quaglin "a menace to society", that his actions were "shocking and lawless", concluding: "What an outrage -- what a disgrace!" Quaglin got 12 years; only 7 other rioters have got longer sentences. Maybe the bottom line here is that Trump and MAGA are steep decline and have no future -- exiting the stage not with a bang, but with a whimper.

AND SO ON: Late-breaking reports are coming in from Ukraine that air-launched Small Diameter [Glide] Bombs being used in combat "to good effect". Since the original SDB is GPS-guided & the Orcs are jamming GPS, we can bet the SDBs in use either have laser guidance, or are the smarter SDB-2s. Ground-launched SDBs may be coming up soon.

I'm always baffled at people online who think they are smarter than the people who are actually fighting the war. Me, I'm just trying to figure out what is actually going on, when there's so little information -- and much of that is actually disinformation. It's interesting to piece together the puzzle, though some of the pieces are secret, and will be a big surprise when they come to light.

* As reported in an article from CNBC.com ("How Biden Is Winning Over Corporate America Behind Closed Doors" by Rebecca Picciotto, 20 May 2024), President Joe Biden has made it clear since he took office that he's willing to take on big business and billionaires -- putting on a show of union support, aggressive antitrust regulation, a crackdown on "junk fees" trying to raise taxes on the wealthy, and blaming corporate greed for inflation. A recent White House declaration called out Big Pharma, grocery chains, credit card companies, airlines, and student debt creditors for "price gouging," one of the Biden Administration's standard lines of attack.

The business community is not happy with being called out, and invariably sues the Biden Administration for every regulatory action it takes. However, Joe has also made it clear that he is not interested in crushing business, instead trying to persuade them to play fair. Behind closed doors, things are not quite so hostile. Neil Bradley, a senior official at the US Chamber of Commerce, suggests that the public tone of the White House sometimes goes "too far", but still says:

QUOTE:

END_QUOTE

Early in May, Biden and his cabinet hosted eight executives at the White House for a private roundtable where they talked about infrastructure investment, geopolitical issues and the US economy's performance relative to the rest of the world. The next day, Biden visited the battleground state of Wisconsin to play up a $3.3 billion USD investment to build an artificial intelligence data center from Microsoft.

Biden has built a pro-labor, pro-consumer economic brand, but he needs the help of industry leaders to make private infrastructure investments, and their support could be useful in the elections in November. Senior White House officials have been calling up business CEOs, while those same CEOs have been invited to the White House to get their views on what to do to achieve the administration's agenda -- while battles between the two sides go on in court.

Biden and the business community know they need each other. Bradley says:

QUOTE:

END_QUOTE

[FRI 24 MAY 24] JOE BIDEN (27)

* JOE BIDEN (27): The crash of the Bork nomination left the Reagan White House looking for another candidate to replace Lewis Powell on SCOTUS. After some gymnastics, they came up with Anthony Kennedy, who was on a California Federal Court of Appeals. Although Reagan himself had talked confrontationally while defending the Bork nomination, the consensus in the White House was finally that a less confrontational approach was needed. Reagan was not comfortable with confrontation anyway.

Reagan asked Joe to come to the White House to discuss the nomination. Despite all the sound and fury over the Bork nomination, Reagan greeted him warmly with: "Hi, Joe!" -- and shaking his hand. Reagan was a very smooth operator, having a way of making people feel welcome. He told Joe: "Congratulations on Bork."

Joe replied: "No, Mr. President. There's no cause of congratulations. I feel bad for Judge Bork. He was a good man."

Reagan countered with: "Ah, he wasn't all that much." The warm greeting was not a surprise to Joe, warmth was natural to Reagan, but the casual dismissal of Bork was startling -- though given Bork's complaints about the White House's failure to push the nomination, maybe shouldn't have been so surprising. It also demonstrated that Reagan, when he suffered a failure, simply swallowed it and moved on -- which had a lesson for Joe's recent failed presidential run.

Anyway, Reagan moved immediately on to other prospects to fill Justice Powell's seat, asking: "Who do you want, Joe?"

Joe replied: "Mr. President, that's not my job ... I'm happy to give you an honest appraisal of the prospects for potential nominees." Reagan then went to give Joe a set of candidates, one by one, with Joe providing a judgement on whether they could pass Senate muster or not. The last on the list was Judge Anthony Kennedy, of the 9th Circuit Court in California. Joe said: "Based on what I know, he's a mainstream conservative. He would probably pass."

"So that means you're for him?"

"No, based on what I know, he'd pass the Senate. I might vote for him, but I don't know enough."

Reagan was only going to take YES for an answer: "Right. You're for him." Reagan was partly pranking Joe, but that conclusion was still obvious: if Joe couldn't think of any reason to oppose the nomination, he was unlikely to find one, and Judge Kennedy was going to get a PASS. Besides, he had Irish roots, which couldn't have displeased Joe. There wasn't much more to say after that, but when Joe got up to leave, Reagan asked him: "Joe, have you got a few more minutes?"

"Of course, Mr. President." Reagan led him to the private study off the Oval Office, where Anthony Kennedy was waiting. Reagan was clearly enjoying pranking Joe, telling Kennedy: "Tony, Joe says he's for you!" Joe tried to explain that wasn't what he said, but it was futile; Reagan had subtly out-maneuvered him. Joe found the "well-rehearsed theater" of the meeting amusing.

Reagan announced Kennedy's nomination on 11 November, making it clear that Kennedy did believe the Constitution granted "unenumerated rights", and that his, Reagan's, earlier confrontational rhetoric had been "facetious". Some civil rights advocates were suspicious of Kennedy's record, but Joe shot back that "no one is fully satisfied with his philosophy or how he would rule -- which means he in fact is as open-minded as I hoped he would be."

Anthony Kennedy sailed through the confirmation and ascended to the SCOTUS bench. He would prove an asset to the court, being the swing vote in many important cases, and Joe would never regret his appointment. [TO BE CONTINUED]

[THU 23 MAY 24] WINGS & WEAPONS

* WINGS & WEAPONS: As discussed in an article from THEDRIVE.com ("Ukraine's Jet Powered UJ-25 Skyline Drone Appears On Battlefield" by Joseph Trevithick, 22 December 2023), the Ukraine War has been dominated by drones, with Ukraine's defense industry generating new designs on a steady basis.

One of the latest, the Ukrjet "UJ-25 Skyline", is something of a departure for Ukraine, being powered by a small turbojet. It recently broke cover when one of them crashed into the roof of a house in the Russian-occupied Ukrainian city of Berdyansk. Details were not provided by Ukrainian sources, but it is clearly a derivative of the Ukrjet "UJ-23 Topaz" -- originally developed as a target drone, adaptable to the "intelligence, surveillance, reconnaissance (ISR)" mission.

Both the UJ-25 and UJ-23 have sleek bodies, low-mounted forward-swept wings, vee tail, and an engine intake in the back. Although Ukrjet has not released details about the UJ-25, the company has provided details of the UJ-23 -- saying it cruises at 500 KPH (370 MPH), has a maximum speed of 800 KPH (500 MPH), an endurance of 90 minutes, and an operational radius of 400 kilometers (250 miles). It is described as "runway independent" -- meaning launch by catapult or rocket booster -- and is recovered by parachute.

The UJ-25's performance is likely similar, giving it a range of at least 800 kilometers (500 miles), enough to penetrate areas of Russia around Ukraine's borders. Nothing is known about its guidance systems or payloads, though it could obviously be used as a strike asset, decoy, or jammer platform. Incidentally, forward-swept wings enhance "stealth", since they deflect radar signals from the front towards the drone's fuselage, instead of back towards a radar. Forward-swept wings tend to be overstressed in aggressive maneuvering, but drones typically don't perform aggressive maneuvers.

* As discussed in an article from THEDRIVE.com ("New Ukrainian Kamikaze Jet Drone Appears In Russia" by Thomas Newdick, 7 February 2024), Ukraine has fielded another jet drone, known only from a "dud" example captured by the Russians. The Ukrainians haven't said anything about it, so it's just being called the "Black Dolphin K-Drone (BDKD)" for its appearance.

It is a sleek black machine about 3 meters (10 feet) long, with a low-mounted ogival delta wing -- bell shape, sharp sweep at the wing root, curving to moderate sweep in mid-wing, then back to a sharp sweep at the tips -- with a tailfin and a jet intake in the belly. Russian sources say it is of wooden construction, with some carbon-fiber parts and fiberglass skin. The smooth contours and non-metallic materials suggest it has some degree of stealth.

The BDKS has a 20-kilogram (44-pound) warhead and is powered by a little P400-PRO turbojet engine from the German JetCat company -- a centrifugal-flow engine, with a single compressor stage like a pump impeller instead of a disk fan, that being a common configuration for small cheap turbojets. The intake on the belly is unusual, engine intakes tending to be unstealthy features, normally placed on top to shield them from radar. The position on the belly suggesting that the BDKS flies at low altitude.

The latest Ukrainian long-range drones do not rely on GPS guidance, instead featuring unjammable schemes like terrain-following. Specifications like speed and range are unknown.

* The Russians have successfully employed their "Lancet" drone in Ukraine. It has cruciform pop-out straight wings and tailfins, a pusher prop, and a range of up to 40 kilometers (25 miles) -- about twice that of a quadcopter kamikaze drone. It appears to be optically guided. The Lancet can be used both for surveillance and as a k-drone.

The Lancet has proven a dangerous nuisance to Ukrainian forces. The Ukrainians decided that they might as well build their own equivalent, with a number of solutions from some of the Ukrainian drone companies. The Ukrainian Lancets look like their Russian counterpart, and have a similar range. Not much has been said of them; it is interesting to wonder if they have options for an electronic-warfare seeker, allowing them to target and attack Russian "emitters" -- radar, radio, or jammer systems.

A photo emerged of yet another Ukrainian k-drone -- of conventional configuration, with straight wings and orthodox tail configuration, plus a pusher propeller. What was unusual about this item was that it was fitted with a standard rocket-propelled grenade (RPG) warhead. That's a very good idea, since RPGs are so very common; such a warhead can't penetrate heavily tank frontal armor, but it can do a good job against relatively light top armor.

[WED 22 MAY 24] GEN-AI ARRIVES (15)

* GEN-AI ARRIVES (15): As discussed in an article from NATURE.com ("ChatGPT Has Entered The Classroom" By Andy Extance, 15 November 2023), it is clear that generative AI is going to have a massive impact, it's just not clear exactly what it will be.

GAI has a particular potential in education. In 2023, educational psychologist Ronald Beghetto -- of Arizona State University (ASU) in Tempe -- asked a group of graduate students and teaching professionals to discuss their work, but not with him. They instead conversed with a set of creativity-focused chatbots Beghetto had put together at ASU.

These bots were based on the same GAI technology that powers the famous and conversationally fluent ChatGPT. Beghetto prompted the bots to take on various personas to encourage creativity -- for example, by deliberately challenging someone's assumptions. One student discussed various dissertation topics with the chatbots, while lecturers talked with the chatbots about how to design classes. Everyone who participated was impressed, one asking: "When are these things going to be available?"

Many educators have been unhappy with the emergence of ChatGPT, thinking it will make it easier for students to cheat on assignments. Beghetto and others exploring the potential of large language models (LLMs) such as ChatGPT are much more optimistic, seeing them as tools to enhance education.

For example, using LLMs to read and summarize large stretches of text could save students and teachers time and help them to focus instead on discussion and learning. ChatGPT's ability to discuss nearly any topic raises the prospect of using LLMs to create a personalized, conversational educational experience. Some educators see them as potential "thought partners", tutors that don't cost as much as a human and are always available. Theodore Gray -- co-founder of Wolfram Research, a technology company in Champaign, Illinois -- says:

QUOTE:

END_QUOTE

Gray says that Wolfram Research is currently working on an LLM-based tutor, but gave no specifics.

Since California firm OpenAI launched ChatGPT in November 2022, much of the attention regarding its use in education has been negative. LLMs work by learning how words and phrases relate to each other from training data containing billions of examples. In response to user prompts, they then produce sentences, including the answer to an assignment question, and even whole essays.

Unlike previous AI systems, ChatGPT's answers are often well written and seemingly well researched. That suggested to the critics that lazy students would just pull their answers out of a chatbot and not really learn anything. Worse, ChatGPT is also notoriously fragile, sometimes giving wildly wrong answers -- even making things up, an issue known as "hallucination".

Wang Wei -- a computer scientist at the University of California, Los Angeles -- found that GPT-3.5 and its successor, GPT-4, were often badly wrong when tested on questions in physics, chemistry, computer science and mathematics taken from university-level textbooks and exams. Wang and her colleagues experimented with different ways to query the two GPT bots. They found that the best method used GPT-4, and that its bot could answer around one-third of the textbook questions correctly that way, although it did score 80% in one exam.

Privacy is another issue: students may not be happy using LLMs once they realize that everything they type into them is being stored by OpenAI and might be used to train the models. [TO BE CONTINUED]

[TUE 21 MAY 24] EUROPEAN FARMERS & CLIMATE CHANGE

* EUROPEAN FARMERS & CLIMATE CHANGE: As discussed in an article from REUTERS.com ("In Climate Fight, Europe's Olive, Wine Farmers Turn To Tech And Tradition" by Charlie Devereux, Antonella Cinelli, & Corina Pons, 14 September 2023), in the third decade of the 21st century, climate change is starting to bite. Farmers in Spain and Italy are suffering painful losses from a warmer planet, and seeking ways to adapt.

Farmers in Spain, for example, are rediscovering the traditional use of crops planted between trees to protect the soil in olive groves, while Italian farmers are using technology to control water usage. Spain and Italy are the world's top producers of olive oil, but the industry is under threat from desertification and drought. Spanish agricultural authorities believe that more than a fifth of its land is at high risk of becoming infertile.

Italy, known for wines and pasta wheat, suffered one of its most severe droughts in 70 years in 2022 and saw another scorching summer in 2023, along with pounding hailstorms in the north. The Italian farming industry is the European Union's third-largest in terms of production value, behind France and Germany but ahead of Spain, in fourth place.

Andrea Ronca, who grows cereals for his family's cattle farm in the province of Mantua in northern Italy, uses satellite images to track where his land is driest: "I can adjust irrigation at any time, even from my smartphone, avoiding any wastage."

Spanish farmer Miguel Moreno was an early adopter of "cover crops". Almost 30 years ago, he began growing grasses alongside trees to stop water draining away down the hill on his olive grove in the southern region of Andalusia. His son, Angel, said that before that large cracks would form where the soil eroded: "You had to be careful because you could fit your foot in them." Cover crops are now used on about 30% of olive groves in Spain.

The changes are being driven both by European Union environmental incentives and reality. In 2022, Spain's drought-hit olive oil production slumped to less than half of what it had been for the previous four harvests, with 2023 expected to be at least as bad. Wine output in Italy is forecast to fall 12% in 2023 as extreme weather and fungal diseases hit vineyards, meaning Italy will lose its position as the world's largest wine producer to France.

Agricultural researchers say restoring cover crops such as grasses and legumes in the rows between olive trees can act as a sponge to conserve water, prevent flash floods, and restore nutrients. Cover crops used to be commonplace, but tilling and clearing vegetation became widespread with the introduction of industrial pesticides and fertilizers in the 1960s and 1970s.

Cover crops may reduce yields over the short term, but pay off over the longer term. Emilio Gonzalez -- a professor of rural engineering at the University of Cordoba -- says cover crops "function as micro-reservoirs that mean each drop of rainwater stays on the ground longer before running off, so that water is more likely to penetrate."

The EU is subsidizing farmers who use cover crops as part of the new "Common Agricultural Policy (CAP)". The practice can restore biodiversity, promoting insects like the green lacewing, that in turn may help farmers control pests such as the olive fruit fly and the olive moth. Olive farmer Angel said that since introducing cover crops he noticed an increase in birds such as owls, blackbirds, turtle doves, hoopoes, and reptiles such as the near-threatened ocellated lizard. Swiss-headquartered agrichemicals and seeds group Syngenta began offering a cover crop seed mixture in Spain in 2023, and expects the practice to boom in 2024.

Cover crops are well-established in Italy. Farmers there are now exploiting digital technology to give them an edge, particularly in water conservation, with agtech startups lending a hand. The share of land farmed using digital tools rose to 8% in 2022, from 6% the year before, while spending rose to 2.1 billion euros, from just 100 million euros in 2017. Professor Marco Trevisan, Dean of the Faculty of Agricultural Sciences at the Catholic University of Piacenza, says that water conservation is increasingly important as Italy dries up: "We were late because we were used to having lots of water at our disposal, especially in the north."

Simone Rech -- who produces about 250,000 bottles a year mainly of Prosecco sparkling wine, from the province of Treviso, not far from Venice -- says: "We are in the hills, where there is little water." In Rech's vineyard, sensors now monitor the air and soil to monitor factors such as temperature and evaporation from the leaves. Wherever possible, rainwater and waste water are collected, purified and reused.

[MON 20 MAY 24] THE WEEK THAT WAS 20

DAYLOG MON 13 MAY 24: Reports are coming in of a Russian offensive across the border in the direction of Kharkiv. The reports are utterly confused, some saying the Russians are on a roll, others that it's no big deal and Kharkiv is in no danger.

This is of course the "fog of war" at work, but it points to the significant influence of disinformation in that "fog". One pessimistic report had a Ukrainian soldier saying that Kharkiv's defenses had been criminally neglected and the Russians were going wild.

Huh? Kharkiv is more vulnerable to a Russian cross-border attack than anywhere else in Ukraine, and they didn't prepare defenses? I'm actually certain they did. It's just that they didn't build them close to the border. It would make no sense to build a defensive line on unfavorable terrain, and defenses right on the border would be vulnerable to a surprise attack. Better to let the Russians penetrate and stick their necks out, vulnerable to destruction themselves in predefined "kill zones".

I'm sure things are difficult for Ukraine right now, but there may be an element of luring the Russians in so they can be destroyed. There's also talk that new supplies of ammunition will take a long time to reach Ukrainian fighters, but that may be disinformation, too.

A recent video showed AFU forces annihilating a Russian armored column with drones and extremely precise artillery fire. It appears new munitions are coming in, not merely in quantity but with new technology that isn't affected by Russian countermeasures.

DAYLOG TUE 14 MAY 24: Trump is in court these days, and making a pest of himself to the extent that the judge fined him for contempt, warning him that he could be locked up if he persisted. Trump then called up his minions to troll the court in his place. Now we are getting the likes of Rick Scott and JD Vance denouncing the "kangaroo court" picking on Trump. Sigh, these people just do not realize they are on a sinking ship. In any case, they inspired me to cook up a take-off on Gilbert & Sullivan:

   I am the very model of a modern MAGA criminal!
   I make no sense, I have no shame, my ethics they are minimal!
   The facts are what I say they are,
   I change them if I like!
   I'm on the side of RT trolls,
   I'm really not that bright!
   I can not be trusted, I'm ignorant and whimsical!
   I am the very model of a modern MAGA criminal!

* Regarding comments yesterday on disinformation and the "fog of war", today it is reported that General Kyrylo Budanov, head of Ukrainian HUR military intelligence describes the situation on the Kharkiv front as "critical", while field commanders are saying it's "under control".

I call Budanov the "Tanuki" -- the Japanese raccoon-dog, in legend a shape-shifter and trickster. What he says has to be assumed to be clever disinformation. He's too smart to lie; the best disinformation is true, just not the entire and balanced truth. A Ukrainian soldier was also saying that it would take a few months to get ammunition supplies up to speed. I suspect it's more like a few weeks.

DAYLOG WED 15 MAY 24: Supreme Court Justice Clarence Thomas has long been controversial, and as of late has become more so. His wife, Ginny Thomas, has been identified as a supporter of Donald Trump's "Clown Car Coup" to overturn the 2020 election, and Clarence Thomas himself has been accused of receiving lavish gifts from wealthy Rightists. The tale is that Thomas had been dropping hints that he might resign from SCOTUS because he wasn't making enough money there.

OK, that looks very bad, but it's mostly news reports that may not be reliable. However, British comedian John Oliver -- of the LAST WEEK TONIGHT show -- has taken them at face value, and with a certain Zen has publicly told Thomas: I have such a deal for you!

Oliver announced on LWT in February: "We have a special offer for you tonight. We are prepared to offer you $1 million a year for the rest of your life, if you simply agree to leave the Supreme Court immediately, and never come back."

Oliver also offered to give Thomas a multi-million luxury RV, presumably a converted tour bus -- that reflecting on Thomas' 1999 purchase of a high-end motor home, with friendly financing help from a rich admirer.

Oliver's offer was good for 30 days. Thomas, to no surprise, didn't bite, but early this month Oliver offered it again. Thomas has been complaining as of late about what a nasty place Washington DC is, so maybe he's more willing to bite now.

As Oliver put it in February: "A million dollars a year and a brand new condo on wheels, and all you have to do in return is sign the contract and get the f*** off the Supreme Court." Oliver isn't into subtle British wit. He more believes in scorched earth.

I'm still fond of Oliver's comment from some years back about the Trump Administration being "Stupid Watergate" -- as bad as the original, but "everyone involved is stupid and bad at everything." That was said while Trump was in office; Oliver was wrong ... it was a lot worse than the original in all respects.

DAYLOG THU 16 MAY 24: As reported by VOX, the US Supreme Court has now ruled on the case of the US Consumer Financial Protection Bureau (CFPB) V. Community Financial Services Association (CFSA) of America. As is typical of SCOTUS cases, it was complicated.

The CFPB was created as a consequence of the Dodd-Frank Wall Street Reform & Consumer Protection Act of 2010, with the CFPB of course being focused on consumer protection -- one aspect being regulation of financial institutions, plus advice to help them follow the law.

CFSA, on its part, is the trade association of the payday-lending industry -- which has a reputation for predatory business practices and high interest rates. The CFSA wasn't happy with CFPB rules so they sued in 2018, with the Western Texas District Court rejecting the claim.

Of course the CFSA didn't give up, taking the case to the infamously reactionary 5th Circuit Court of appeals -- which declared the CFPB to be unconstitutional, because it has an indirect funding path via the Federal Reserve. The case went up to SCOTUS.

There was a lot at stake: killing off CFPB would throw America's mortgage system into chaos. SCOTUS ruled 7:2 in favor of CFPB, with Justice Clarence Thomas writing the decision for the majority. Even Thomas thought the case was bunkum. Justices Samuel Alito and Neil Gorsuch dissented, it seems complaining about the indirect funding path through the Fed. The majority replied, in effect: So what? The Constitution is a brief document; it doesn't nail everything down, and it's foolish to pretend it does.

DAYLOG FRI 17 MAY 24: I was saying a few days ago that tales of weak Ukrainian defenses north of Kharkiv sounded like disinformation. UKRAINSKA PRAVDA has now run a story with Zelenskyy saying there were three lines of defense, and the Russians haven't reached the first one.

Zelenskyy also confirmed that it wouldn't make any sense to put the defenses right on the border, because the Russians would always be bombarding them and killing Ukrainian soldiers. In addition, Zelenskyy says the troops are no longer short of artillery ammunition, though air defense is still in a difficult state.

I find it strange that so few people seem to understand that war means disinformation. As of late, I'm particularly wondering if every announcement of a new critical failure in Ukraine's defense -- leaky anti-air capabilities as of late -- means new weapons are coming in to fix the problem. That's how disinformation works: if I'm confused, so are the Orcs.

I'm also suspecting that this offensive is Putin's last big throw. He wants to break Ukraine before the buildup of arms and the depletion of his resources makes it impossible to do. I have to be cautious, since in 2023 I underestimated Putin, not realizing that he would mobilize Russia to such a degree to pursue his war. However, Putin can't sustain his war effort indefinitely, and the odds are shifting against him.

* I'm using the DeviantArt DreamUp image generator these days ... nothing risque, all "safe for viewing". I focus on glam and I have female followers -- some of them may be DC, of course, but they may just like the styles.

I did Joe Biden as "Joe B-Wan Kenobi"; it took a lot of tries, but I finally got something that worked. DreamUp still does a bad job drawing hands, and can come up with weird hallucinations like three legs or the like. Hands are actually very complicated, humans have a hard time drawing them. For now, DreamUp tends to hide hands, putting them in pockets or such. I wish they'd fix that stuff.

Anyway, when I posted the Joe B-Wan Kenobi image to DeviantArt, to my surprise I got a MAGA troll bad-mouthing JB. When I was on Xitter I saw trolls all the time, but I'm not used to them any more. I just blocked him. I was a little apprehensive that the trolling would ramp up -- but it didn't. I don't think trolling gets much traction on DeviantArt. In any case, trolling is in decline.

AND SO ON: In news of the Gaza Crisis, it is now being reported that Benny Gantz -- a member of Israel's War Cabinet and Prime Minister Benjamin Netanyahu's main political rival -- said he would leave the government on 8 June if it did not come up with a new war plan that sets up an international, Arab, and Palestinian administration to handle civilian affairs in Gaza.

Defense Minister Yoav Gallant, another member of the Cabinet, has also called for a plan for Palestinian administration, and said in a speech that he wouldn't agree to Israel's continued occupation of Gaza. The USA, on its part, has called for a revitalized Palestinian Authority to govern Gaza with assistance from Saudi Arabia and other Arab states, on a path towards eventual statehood. US National Security Advisor Jake Sullivan is headed to Israel to push those plans.

In response to Gantz's ultimatum, Netanyahu said his conditions would amount to "defeat for Israel, abandoning most of the hostages, leaving Hamas intact, and establishing a Palestinian state." If Gantz pulls out, Netanyahu will be even more dependent on hard-liners who want Israel to reoccupy Gaza, encourage the "voluntary emigration" of Palestinians from the territory, and reestablish Jewish settlements that were removed in 2005.

Polls show that Netanyahu would be driven from office if new elections were held, with Gantz most likely to replace him. That would likely mark the end of Netanyahu's long political career and expose him to prosecution over longstanding corruption charges. Although it's hard to be optimistic when it comes to Israel versus the Palestinians, it does seem like there's finally an opportunity to dramatically alter the playing board -- but Netanyahu will have to go first.

[FRI 17 MAY 24] JOE BIDEN (26)

* JOE BIDEN (26): On the morning of Wednesday, 23 September 1987, Joe drove to Washington DC, accompanied by Jill and two aides. The star of the hearings that morning was retired Chief Justice Warren Burger -- a Republican, Center-Right, liberal in some ways, staunchly conservative in others. After discussion, Joe got down to the heart of the matter with Burger:

QUOTE:

END_QUOTE

Burger replied:

QUOTE:

END_QUOTE

Joe was startled at that claim, since Bork clearly did disagree with it. Joe went on with Burger, asking him if he affirmed the 9th Amendment, with Burger answering somewhat vaguely. Joe left it at that and yielded to Strom Thurmond.

The hearings dragged on into a third week, but Bork's chances were obviously fading. On 6 October, the Judiciary Committee took a vote, with nine voting against confirming Bork, five voting for confirming him. Arlen Specter of Pennsylvania was the only Republican to vote NAY; he was notoriously independent, having been a Democrat to 1965, and would much later become one again.

That didn't end the proceedings, with the Bork nomination then moving towards a full Senate vote. Bork and the Reagan White House went all-out to save the nomination, but it couldn't be saved: the Senate voted on it on 23 October, with the Bork nomination rejected by 58:42, an overwhelming margin. The conclusion of the Bork hearings eased the sting of the failure of Joe's presidential campaign, and also help establish that Joe was not an intellectual lightweight. However, it was Bork who, again, had hanged himself, Joe much later saying that "his arrogance was his undoing."

Bork was of course resentful of the way the hearings had turned out -- saying they couldn't be called "hearings", because nobody heard what he was trying to say. He did not seem to understand that his nuanced read on his judicial views simply looked like tap-dancing around vital issues. He was not so unhappy with Joe, complaining more loudly about the failure of the Reagan White House to push through the nomination. Bork went back to the DC Court of Appeals, but resigned a few months later. Financially, the hearings left him in good shape, with a big book advance and lucrative speaking engagements. He ended as a high-profile legal scholar, being associated with Rightist think-tanks. [TO BE CONTINUED]

[THU 16 MAY 24] SPACE NEWS

* Space launches for April included:

[02 APR 24] USA VB / FALCON 9 / STARLINK 7-18 -- A SpaceX Falcon 9 booster was launched from Vandenberg SFB at 0230 UTC (old day local time + 7) to put 22 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[02 APR 24] CN XC / LONG MARCH 2D / YAOGAN-42 01 -- A Long March 2C booster was launched from Xichang at 2256 UTC (next day local time - 8) to put a secret "Yaogan-42 01" payload into orbit. It was apparently a military surveillance satellite.

[05 APR 24] USA CC / FALCON 9 / STARLINK 6-47 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 0912 UTC (local time + 4) to put 23 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[07 APR 24] USA VB / FALCON 9 / STARLINK 8-1 -- A SpaceX Falcon 9 booster was launched from Vandenberg SFB at 0225 UTC (previous day local time + 7) to put 22 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[07 APR 24] USA CC / FALCON 9 / BANDWAGON 1 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 0212 UTC (previous day local time + 4) on the "Bandwagon 1" mission, to put 11 smallsat payloads into space. The payloads included:

• Project 425 2: SAR satellite for South Korea's DAPA, part of a constellation of five surveillance satellites.
  
  
• Acadia 4 (Capella 14): Another SAR microsat from Capella Space.
  
  
• Hawk 8A,B,C;9A,B,C: Six more SIGINT satellites from HawkEye 360.
  
  
• QPS-SAR 7 (TSUKUYOMI II): An Earth observation satellite from iQPS of Japan.
  
  
• TSAT 1A: An Earth observation satellite from Tata Advanced Systems LTD (TASL) of India.
  
  
• TBA (425 Satellite): Surveillance satellite from South Korea's DAPA.
  
  
• Centauri 6: A 12U IoT CubeSat from Fleet Space of Australia.

[09 APR 23] USA CC / DELTA 4 HEAVY / NROL 70 (USA 353) -- A Delta 4 Heavy booster was launched from Cape Canaveral at 1653 UTC (local time + 4) to a classified National Reconnaissance Office payload into orbit. The payload was designated "NROL 70 (USA 353)". It was believed to be an ORION / MENTOR-type SIGINT satellite.

[10 APR 24] USA CC / FALCON 9 / STARLINK 6-48 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 0514 UTC (local time + 4) to put 23 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[11 APR 24] RU-C VT / ANGARA A5 ORION / GAGARINETS -- An Angara A5 Orion booster was launched from Vostochny at 0900 UTC (local time - 8) to put the "Gagarinets" satellite into geostationary orbit.

[11 APR 24] USA VB / FALCON 9 / USSF 62 -- A SpaceX Falcon 9 booster was launched from Vandenberg SFB at 1200 UTC (local time + 7) on the "USSF 62" mission. Payloads included:

• Weather System Follow-on-Microwave 1 (WSF-M 1): A USSF space weather satellite. Developed by Ball Aerospace, the satellite carried a microwave imaging radiometer and an energetically charged particle sensor space weather payload.
  
  
• Athena: An Air Force satellite, built by NovaWurks.
  
  
• TYRAD 1,2: Two satellites from Auburn University in Alabama, intended for gamma-ray astronomy.

[13 APR 24] USA CC / FALCON 9 / STARLINK 6-49 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 0122 UTC (previous day local time + 4) to put 23 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[15 APR 24] CN JQ / LONG MARCH 2D / SUPERVIEW NEO-3 1 -- A Long March 2C booster was launched from Jiuquan at 0412 UTC (local time - 8) to put the "SuperView Neo-3 01 (Siwei Gaojing-3 1)" satellite into orbit for China Siwei.

[17 APR 24] USA CC / FALCON 9 / STARLINK 6-51 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 2126 UTC (local time + 4) to put 23 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[18 APR 24] USA CC / FALCON 9 / STARLINK 6-52 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 2240 UTC (local time + 4) to put 23 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[20 APR 24] CN XC / LONG MARCH 2D / YAOGAN-42 02 -- A Long March 2C booster was launched from Xichang at 2345 UTC (next day local time - 8) to put a secret "Yaogan-42 02" payload into orbit. It was apparently a military surveillance satellite.

[23 APR 24] USA CC / FALCON 9 / STARLINK 6-53 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 2217 UTC (local time + 4) to put 23 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[23 APR 24] NZ / ELECTRON / NEONSAT 1, ACS3 -- A Rocket Labs Electron light booster was launched from New Zealand's Mahia Peninsula at 2232 UTC (next day local time - 13) to put the South Korean "New space Earth Observation Satellite (NeonSat) 1" and NASA's "Advanced Composite Solar Sail System (ACS3)" CubeSat into orbit.

NeonSat 1 was an imaging surveillance microsatellite with a launch weight of about 100 kilograms (220 pounds), apparently for both military and civil use. It was developed by the Satellite Technology Research Center (SaTReC) at the Korea Advanced Institute of Science and Technology (KAIST), Korea's leading science and technology institute. It was a pathfinder for a constellation of ten more NeonSats, all to be launched by 2027.

ACS3 was based on a 12-unit CubeSat. It deployed a square one-piece solar sail 9 meters (30 feet) on a side, using four booms made of composite material.

[24 APR 24] CN JQ / LONG MARCH 2F / SHENZHOU 18 -- A Long March 2F booster was launched from Jiuquan at 1258 UTC (local time - 8) to put the "Shenzhou 18" crewed space capsule into space on a Tiangong China Space Station support mission. The three-taikonaut crew included commander Ye Guangfu, operator Li Cong, and system operator Li Guangsu.

[28 APR 24] USA CC / FALCON 9 / GALILEO FM25 & FM27 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 0034 UTC (previous day local time + 4) to put the "Galileo FM25 & FM27" navigation satellites into orbit.

[28 APR 24] USA CC / FALCON 9 / STARLINK 6-54 -- A SpaceX Falcon 9 booster was launched from Cape Canaveral at 2208 UTC (local time + 4) to put 23 SpaceX "Starlink v2 Mini" low-Earth-orbit broadband comsats into orbit.

[WED 15 MAY 24] GEN-AI ARRIVES (14)

* GEN-AI ARRIVES (14): AI image-generation tools can also benefit from automatically generated prompts. A team at Intel labs, led by Vasudev Lal, set out to optimize prompts for the image-generation model Stable Diffusion. Lal saw the exasperating finickiness of writing prompts as a bug, saying: "So we wanted to see if we can automate this kind of prompt engineering."

Lal's team built a tool named "NeuroPrompts" that takes a simple input prompt, such as "boy on a horse," and automatically tweaks it to produce a better picture. To do this, they started with a range of prompts generated by human prompt-engineering experts, and then trained a language model to transform simple prompts into these expert-level prompts. On top of that, they used reinforcement learning to optimize these prompts to create more aesthetically pleasing images -- as rated by yet another machine-learning model, PickScore, a recently developed image-evaluation tool.

The automatically generated prompts did better than the expert-human prompts they used as a starting point, as rated by the PickScore metric. Lal wasn't surprised: "Humans will only do it with trial and error. But now we have this full machinery, the full loop that's completed with this reinforcement learning ... This is why we are able to outperform human prompt engineering."

Of course, the automatic approach still doesn't necessarily give users what they want, so they gave uses the ability to guide the automated process. In their tool, the user could specify the original prompt -- like "boy on a horse" -- and an artist to emulate; or a style; or a format; or other modifiers.

Lal believes that the exasperating finickiness of LLMs is temporary, a sign of the infancy of the technology, saying: "I think it's important that these kinds of optimizations are investigated and then ultimately ... incorporated into the base model itself so that you don't really need a complicated prompt-engineering step." It's just a question of training AI to be smarter about handling the prompts it's handed.

Tim Cramer -- senior vice president of software engineering at Red Hat -- suggests that automating prompt generation makes sense for casual mass-market applications, but doesn't work so well for more specialized applications. Cramer says: "I think there are going to be prompt engineers for quite some time, and data scientists. It's not just asking questions of the LLM and making sure that the answer looks good ... there's a raft of things that prompt engineers really need to be able to do."

Austin Henley says there are many criteria to deal with in using LLMs to produce commercial product:

• Ensuring reliability -- that is, not failing easily, and failing gracefully when it comes to that.
  
  
• Adapting an LLM's output to the required format.
  
  
• Testing to make sure the AI assistant won't do something bad in even a small number of cases.
  
  
• Ensuring safety, privacy, and compliance. Henley says that testing and compliance for LLMs is particularly difficult, since traditional software-development testing strategies are almost useless for nondeterministic LLMs.

In other words, getting the right use out of an LLM really is a serious engineering task. Many big companies now offer a new job title: "Large Language Model Operations (LLMOp)" -- for which prompt engineering is only a part, along with establishing conformance and testing. Given that GAI is evolving rapidly, any such job is going to evolve just as rapidly. Cramer says: "The landscape is just too crazy right now. Everything's changing so much. We're not going to figure it all out in a few months."

[ED: GAI is sort of where personal computing was in the early 1980s: a lot of interesting things going on, a lot of ferment, with rapidly evolving technology that often appears appallingly crude in hindsight. GAI is going to be working vastly better in 2030 than it does now.

In working with DeviantArt's DreamUp image generator, it can be frustrating to specify prompts that give me what I want. One thing I found out is that I have to specify major features of an image before I specify minor details, working down a hierarchy. It appears that if I specify a minor detail first, DreamUp will assume the major features and ignore my specification for them. Again, sometimes the simplest and shortest prompts will get the best results, but it's kind of pot luck.

DreamUp users can list their prompts along with their imagery, but there's a split mindset: some people, particularly those trying to make money off their "deviations", hide their prompts -- while those with a more tinkering mindset like to share them, and find the protective mindset exasperating.] [TO BE CONTINUED]

[TUE 14 MAY 24] SYNTHETIC DNA CRACKDOWN

* SYNTHETIC DNA CRACKDOWN: As discussed in an article from WIRED.com ("The US Is Cracking Down on Synthetic DNA" by Emily Mullin, 6 May 2024), work on made-to-spec "synthetic DNA" has evolved into an emerging industry. Synthetic DNA has a wide range of application, being used to develop diagnostic tests, make enzymes to digest plastic, or engineer antibodies to treat disease.

A research group needs to investigate a rare species of bacteria? Instead of finding samples, the researchers can just get the sequence from a DNA synthesis company. Synthesizing DNA is nothing all that new, but as of late it's easier, cheaper, and faster to do, thanks to new technology that can "print" custom gene sequences. There are dozens of companies now producing synthetic DNA in volume.

Synthetic DNA has great promise, but it also presents hazards. It's becoming possible to create entirely new sequences that don't exist in nature -- and, by accident or intent, they could include some that present a threat to humans or other organisms. Tom Inglesby -- an epidemiologist and director of the Johns Hopkins Center for Health Security -- says:

QUOTE:

END_QUOTE

It's possible that a Black Hat could make a dangerous virus from scratch by ordering its genetic building blocks and assembling them into a whole pathogen. In 2017, Canadian researchers announced they had reconstructed the extinct horsepox virus for $100,000 USD using mail-order DNA. What if somebody did the same for smallpox, which is extinct in the wild?

On 29 April 2024, the White House issued new rules to regulate at companies that manufacture synthetic DNA. The rules are consequent to an executive order signed by President Joe Biden in the fall of 2023 to establish new standards for AI safety and security, including AI applied to biotechnology. The rules ask DNA manufacturers to screen purchase orders to flag "sequences of concern" -- that might, for example, increase the toxicity of an organism, or enhance its ability to cause disease -- and assess customer legitimacy. At present, the rules only apply to researchers that receive Federal funding: They must order synthetic nucleic acids from providers that implement these practices.

Inglesby says it's still a "big step forward", since about three-quarters of the US customer base for synthetic DNA are federally funded entities. Many synthetic DNA companies already follow screening guidelines issued by the Department of Health & Human Services in 2010. About 80% of the industry has joined the International Gene Synthesis Consortium, which pledges to vet orders. However, these measures are both voluntary, and not all companies comply.

Kevin Flyangolts -- founder and CEO of New York-based Aclid, a company that offers screening software to DNA providers -- says he's happy to see the new rules: "While the industry has done a pretty good job of putting some protocols in place, it's by and large not consistent," he says. He still hopes Congress will pass a law requiring all DNA providers to screen orders. In 2023, a bipartisan group of legislators introduced the "Securing Gene Synthesis Act" to mandate screening more broadly, but the bill has yet to advance.

Emily Leproust -- CEO of Twist Bioscience, a San Francisco DNA-synthesis company -- also wants regulation:

QUOTE:

END_QUOTE

Twist has been screening sequences and customers since 2016, when it first started selling synthetic DNA. A few years ago, the company set up a "red team" of outside consultants to test their screening, with the consultants setting up fake customer names and ordered sequences of concern. The screening caught many of the probes, but there was some confusion on what sequences were "concerning" -- leading to a review and tightened procedures.

Leproust still worries about certain hypothetical scenarios that are beyond her control. For instance, a state actor with bad intentions could start making its own gene sequences: "Probably the biggest risk is if a state wants to build their own DNA synthesis capabilities. They may be able to do it, because states have vast resources."

[MON 13 MAY 24] THE WEEK THAT WAS 19

DAYLOG MON 06 MAY 24: I'm a geezer who plays videogames -- generally arcade games, puzzle games, racing games, since I don't have too much spare time to play games. I'd been looking around for a free version of PAC-MAN to play, but I couldn't find one that worked for me.

Today, I saw an article on an elaborate version of PAC-MAN that had been opened up for public play. It looked fun, but it was a multiplayer game, and I wanted a single-player game. I searched for alternatives, and found a YouTube video on free versions of PAC-MAN. One, OPEN GREEDY, caught my eye. GREEDY was made by French enthusiasts, starting out on DOS it seems, to work up to an open-source version for Windows about ten years ago. I downloaded it, and it turned out to be a nice little piece of work.

One fun feature is that the ghosts let out a "blood-curdling scream" when I eat them. Incidentally, Kroger supermarkets has a promotional HTML5 derivative of PAC-MAN named HUNGRY CART, with a munching shopping cart and ads to gobble up. It's cute, but not very sophisticated.

DAYLOG TUE 07 MAY 24: Today, President Joe Biden spoke on the occasion of Holocaust Memorial Day, condemning Palestinian Hamas terrorists for their savage 7 October 2023 attack on Israel, and reaffirming his support of that country. JB also denounced the "ferocious surge" in antisemitism on college campuses and elsewhere in the wake of the Gaza crisis. The speech did not endear him to pro-Palestinian activists -- but a choice between the Palestinians and the Israelis was not really a choice at all.

JB cannot derail the Israeli assault on Gaza, but he is doing all he can to reduce its impact on Palestinian civilians. For myself, I'm just waiting for the crisis to blow over. It is a distraction from other things, particularly the Ukraine War. The struggle between Palestinians and Israelis is endless. The Gaza crisis will pass and the fighting will die down, to drop off the headlines. At that point, the long-term solution will become the issue ... but nobody has any good idea of what the long-term solution could be.

DAYLOG WED 08 MAY 24: There was a wave of elections in the UK recently, with Labour winning handily over the Tories. To rub in the humiliation, a Tory MP named Natalie Elphicke then crossed the aisle and joined Labour.

Britain's Tories are imploding; my British penpal sees them having maybe 20 MPs after the next general election. Roughly the same can be said here in the USA. The "crisis of conservatism" is transcending borders. How did it happen?

On consideration, America's Republicans made four great mistakes:

1: They drank the anti-government kool-aid, and decided they didn't want to govern any more.

2: They defined themselves by what they were against, not what they were for, working to obstruct instead of getting anything done.

3: They decided to make up their own facts, and change them as convenient.

4: They decided to make up their own rules, and change them as convenient.

What happens after the fall? One scenario is that the GOP splits into an Extreme-Right party, which will slowly die out, and a Center-Right party -- but the Center has shifted Left, and so the Center-Right has ceased to exist.

After all, DEI, LGBT and reproductive rights, climate-change action, sensible gun-safety laws, and in general responsible government are now Center positions. The No Labels group wanted a Centrist candidate, never understanding: We've already got one. His name is Joe Biden.

DAYLOG THU 09 MAY 24: ESSANEWS.com reported on the activities of Ukrainian Atesh partisans in Crimea. Precisely what they are doing is not entirely clear, but they are claiming many casualties among Russian troops and destruction of Russian equipment.

They also report the Russians are transporting many wounded and dead back from the front lines. Ukrainian guerrillas appear to mostly focus on providing intelligence via smartphone on Russian operations, with the AFU reacting with missile or drone strikes.

The guerrillas also perform sabotage, and like to assassinate Ukrainian collaborating officials. Overall, the war is difficult for the Ukrainians right now, with the Russians putting AFU forces under severe pressure in the direction of Kharkiv.

Now that munitions are starting to flow to Ukraine again, odds are increasingly better that the Russian push will be stopped. Once it is, it's hard to think the Russians will be able to do better in the future. We should have a clearer idea of where things are going by the end of the month. Ukrainian and Allied war production is slowly ramping up towards a critical mass. New and possibly decisive weapons are being introduced. For the time being, however, it's still touch-&-go.

DAYLOG FRI 10 MAY 24: An interview was recently conducted with Jack Dorsey, who had been a prime mover in getting Twitter going, to ultimately sell it to Elon Musk. Dorsey has since been associated with a supposed Twitter follow-on named "BlueSky", but now he's distancing himself. Dorsey is no longer on the board of directors of BlueSky, saying that it has gone wrong in the same way that Twitter went wrong. Dorsey, however, is not reflecting on the way Twitter turned into a feedlot for trolls, but on its centralized control.

It seems Dorsey is after a decentralized, distributed social media system open to completely free expression. This sounds like a triumph of idealism over reality, since the #1 problem with social media is actually its weaponization by trolls.

The big issue in social media, whether Dorsey realizes it or not, is that Twitter -- now known as "Xitter", since Musk renamed it "X" -- is dying, and that something will replace it. Inspections of BlueSky and Instagram's Threads show they leave something to be desired. Mastodon seems more viable, but it's more along the lines of the old USENET than Twitter. That leaves Spoutible as an active candidate, since it is much better engineered than BlueSky and Threads, and far better moderated.

The latest feature in Spout, "Accuracy Alerts", allow users to invoke AI-generated fact checks on postings. The trick here is that regulators, particularly in the EU, are big on content moderation, and Spout is doing everything they want. Eventually, governments will have to dump Xitter, and Spout is likely to be the beneficiary. Not incidentally, there was once upon a time an active "Black Twitter", but Musk killed it off. It seems to have obtained a new home on Spout -- though there is no "Black Spout", it's all chocolate-flavored.

AND SO ON: An online commenter named Joe Khalil made some interesting observations:

COMMENT:

END_COMMENT

I replied that if such things had been said in a sci-fi novel written in the 1960s and set in 2024, everyone would have thought it was a comedy.

* An article from PROPUBLICA.com ("IRS Audit of Trump Could Cost Former President More Than $100 Million" by Paul Kiel and Russ Buettner, 11 May 2024), indicates that Donald Trump may be in big trouble with the Internal Revenue Service, concerning his Trump Tower in Chicago:

QUOTE:

END_QUOTE

The IRS let that pass; the agency was badly underfunded and didn't want to go after high rollers who would fight back. However ...

QUOTE:

END_QUOTE

That got the attention of the IRS, with Trump now facing a charge of up to $100 million USD. Since the IRS doesn't discuss its investigations with the public and the situation is complicated to begin with, it's hard to trace what happened and what's going on -- but one Walter Schwidetzky, a law professor at the University of Baltimore and an expert on partnership taxation, told the authors of the article: "I think he ripped off the tax system."

Exactly what happens remains to be seen, but I've been hoping for years that Trump would get nailed on tax charges -- he's obviously been ripping off the IRS for a long time, he's boasted about it. I was hoping he'd go to court, but this works, too.

[FRI 10 MAY 24] JOE BIDEN (25)

* JOE BIDEN (25): Other members of the Senate Judiciary Committee went on to cross swords with Judge Bork. In the meantime, staffers pulled Joe out of the meeting to give him bad news: reporters had tracked down his flotch at the University of Syracuse Law School 22 years earlier, when he had been sloppy in his attributions in a paper. On the face of it, that didn't seem too disastrous, because he'd been cleared of wrongdoing in the incident -- but on consideration, Joe knew the reporters were not going to be nice to him.

He went back to the hearing, but later called a quick meeting with his colleagues on the Judiciary Committee. Joe explained what had happened and offered to step aside as chairman of the nomination hearing. There was a silence for a moment, leaving him with a sinking feeling -- but then the ranking Republican, Strom Thurmond of South Carolina, said emphatically: "Absolutely not. You're my chairman -- you're my chairman."

Ted Kennedy added: "This is ridiculous. You don't need to step aside."

"Well listen, at least let me explain --"

Alan Simpson, a Republican from Wyoming, cut him off: "You don't have to explain anything. We know you." Nobody on the Judiciary Committee wanted Joe to quit. That reassured Joe somewhat, but the report from Syracuse was still hanging over his head. He talked it over with staff, and the next morning called a press conference to explain that he had made a "stupid mistake", concluding: "I was wrong, but did not intentionally move to mislead anybody. And I didn't."

The headlines in the news media in response were of the style: "Biden Admits To Plagiarism". The feeding frenzy simply ramped up, with Joe being glad to get back to the hearings and ignore the noise for a little while. However, he couldn't make the noise go away, and his family was being hounded by reporters. A week after the hearings started, Joe called a meeting of his family in the evening, including his professional aides, to consider dropping out of the presidential race. He was distraught and so were many of the group, Beau saying: "You'll change, Dad. You'll never be the same."

Finally, after going around in this vein, Joe noticed Jean hadn't been saying anything. He asked her: "What do you think, Mom?" She replied: "I think it's time to get out." That effectively decided the discussion, the only thing left being to determine how to manage the exit.

The next morning -- Tuesday, 22 September -- Joe, with Jill on his arm, spoke to reporters outside the hearing room, telling them:

QUOTE:

END_QUOTE

After that was over, Joe felt physically leaden, surprised by how painful the experience was. Ted Kennedy and Jill encouraged him to get back into the hearings, Jill telling him: "You have to win this thing!" He went in and regained control of the hearings, with Kennedy then slipping him a note saying: "There is life after a presidential campaign." Joe could try again in the future. Fortunately, he had no idea of how long he would have to persist.

Joe and Jill took the train back to Wilmington that evening. Jill suggested they eat out that evening, but said he didn't feel like it. Jill said firmly: "We're going out to dinner tonight."

Yes, Ma'am. They went to Restorante Attilio, where they often ate; they got there late, and it was crowded. When they went in, Joe could hear the murmur: "There's Senator Biden." "There's Senator Biden." Joe had been dreading being recognized -- but then one fellow started clapping, and it turned into a standing ovation. [TO BE CONTINUED]

[THU 09 MAY 24] GIMMICKS & GADGETS

* GIMMICKS & GADGETS: As discussed in an article from YAHOO.com ("Company Unveils Bladeless Honeycomb Wind Turbines" by Allison Hunt, 18 December 2023), wind power doesn't scale well at the lower end, small wind turbines not being cost-effective.

Katrick Technologies of Glasgow, Scotland, has a better idea for "micro wind power", based on assemblies of wind power units that can be stacked together. Each element looks like a hexagonal box with no top or bottom, divided into six triangular cells, and lying on a side. Inside each cell are two airfoils, one front and one back, that oscillate up and down in the wind to generate electricity. The hexagonal units can be stacked up in a "honeycomb" as needed.

The approach is cheaper and more reliable than using small turbines, though how cost-effective it is can be argued. It isn't really intended to compete with turbines, being targeted for use in built-up urban areas where obtrusive turbines just don't work.

* In personalized gimmick news, a writeup on the DJI Osmo Pocket 3 camera caught my eye. It's a "vlogging" camera, tuned for wandering around and taking videos, in a convenient one-handed fashion, with the imager on a gimbal mount. Reviews suggested its capabilities approach that of a semi-pro mirrorless camera, though it has one drawback: it only has 2x digital zoom.

My interest in it is for low-light / night still photography. My Samsung Galaxy S20 Ultra does pretty well for low-light photography, but it has painfully limited image stabilization. Videos taken with the Pocket 3 for a night walk down a city street give very good results, even though the shooter was in motion. Lack of zoom is not a big deal for night shots. With a mini-tripod and a large micro-SD card, it should cost about $570 USD, but I will spring for it. I don't buy many spendy things these days and I'm not short of money -- so why not?

* As discussed in an article from NEWATLAS.com ("Climbing Crane Shimmies Up Turbine Towers" by Loz Blain, 2 November 2023), putting up or maintaining tall wind turbines requires a tall crane -- and as the turbines get taller, so must the cranes. The giant turbines now being built mean impractically tall cranes. To get around that problem, German wind turbine company Enercom has collaborated with Dutch company Lagerway to design the "LCC140 Climbing Crane", which crawls up the turbine tower as it's being assembled. Climbing cranes are nothing new, but they're getting new life in the era of wind turbines.

The LCC140 is huge itself, with a weight of 270 tonnes (300 tons); it has to be transported in 11 trucks. It has to be set up on a turbine tower with a separate crane, and towers need to be designed to accommodate it, with reinforced holes in each section of the turbine tower as connection points. It can be set up or torn down in one day. Once set up, as it crawls up the tower it can hoist loads of up to 140 tonnes (155 tons) with its own 33-meter (108-foot) mast and 46-meter (151-foot) boom.

Spanish company KoalaLifter is also working on a turbine-climbing crane, the "KL-N30", but it's smaller, only able to lift 30 tonnes (33 tons), can be hauled by a single truck, and can work on turbine towers of various sizes that haven't been specifically designed to accommodate it. It uses straps to attach to a tower and has two sections, with one section strapped in place while the other moves up -- to strap in place in turn, with the other section moving up.

KoalaLifter is also working on bigger climbing cranes, including the "KL-N80" and "KL-Gorilla" cranes, with load capacities of 80 and 150 tonnes (88 and 165 tonnes) respectively. They will still only need to be hauled by one truck.

[WED 08 MAY 24] GEN-AI ARRIVES (13)

* GEN-AI ARRIVES (13): As discussed in an article from SPECTRUM.IEEE.org ("AI Prompt Engineering Is Dead, Long Live AI Prompt Engineering" by Dina Genkina, 6 March 2024), the arrival of generative AI has presented users with a challenge: figuring out how to phrase a "prompt", or query, to a large language model. It's similar to the art of phrasing queries to Google that find the information we want -- but given that LLMs provide much more detailed information, it's much more difficult. The computer what we tell it to do, which can lead to bizarre results in GAI image / video generation, and useless results in other applications.

Companies are now using GAI to build product copilots, automate tedious work, create personal assistants, and a lot more, according to Austin Henley, a former Microsoft staffer who conducted a series of interviews with people developing LLM-powered copilots. Henley says: "Every business is trying to use it for virtually every use case that they can imagine."

Devising a useful prompt is tricky in the first place, and tends to differ from application to application. A cottage industry of sorts has arisen to provide hints and how-tos on writing prompts, being glorified as "prompt engineering". In reality, prompt design is more art than science, and a dark art at that.

Rick Battle and Teja Gollapudi at California-based cloud computing company VMware were startled at how finicky and unpredictable LLM performance was in response to strange prompting techniques. For example, people have found that asking models to explain its reasoning step-by-step -- a technique called "chain-of-thought" -- improved the performance of the models on a range of math and logic questions. Even stranger, Battle found that giving a model positive prompts, such as: "this will be fun" -- or: "you are as smart as chatGPT" -- sometimes improved performance.

Battle and Gollapudi decided to systematically test how different prompt-engineering strategies impacted an LLM's ability to solve grade-school math questions. They tested three different open-source language models with 60 different prompt combinations each, and found the results wildly inconsistent. Chain-of-thought prompting, for example, sometimes helped and other times hurt performance. They say: "The only real trend may be no trend. What's best for any given model, dataset, and prompting strategy is likely to be specific to the particular combination at hand."

There may be, at least conceptually, a simple way out of the trap: ask LLMs to generate their own optimal prompt. New tools have been developed to that end. Given a few examples and a quantitative success metric, these tools zero in on the optimal phrase to feed into the LLM. Battle and his collaborators have tested these prompt-generation systems and found their prompts almost always did better than the best prompt found through trial-and-error, and found them much faster.

Battle found that some of the automatically-generated prompts were so bizarre that humans were unlikely to ever think of them. Battle says: "I literally could not believe some of the stuff that [one of the systems] generated." He noted one prompt that was straight out of STAR TREK:

QUOTE:

END_QUOTE

It seems that talking to this particular LLM as if it were Captain James T. Kirk improved its skills on math questions. As strange as such results are, Battle thinks that we can't ever figure out better prompts than a machine can. [TO BE CONTINUED]

[TUE 07 MAY 24] RUSSIAN WAR ECONOMY

* RUSSIAN WAR ECONOMY: As discussed in an article from BBC.com ("Russia's War Economy Cannot Last But Has Bought Time" by Faisal Islam, 10 February 24), at the outset of the Ukraine War, Russia was hit with painful sanctions. In March 2022 the Russian rouble had crashed, the value in London of corporate giants Gazprom and Sberbank fell 97%. Queues began to form at cash machines in Moscow. Oligarchs' yachts, football teams, mansions, and even their credit cards were seized. Russia crashed into a major recession.

The toughest sanctions were the seizing of the Russian state's official foreign exchange assets and, in particular, the unprecedented freezing of the central bank's $300 billion USD in currency reserves. Western governments avoided using phrases like "economic warfare", but there was no other good way to describe it.

About two years on, Russian President Vladimir Putin gloated that Russia was the fastest growing economy in Europe, with the International Monetary Fund (IMF) upgrading its forecast growth for 2024 to 2.6% from 1.1%. Putin has mobilized Russia's economy to support the war in Ukraine. Western leaders claim this model is completely unsustainable even over the medium term. The question remains: just how long it can it be sustained?

Russian state spending is at record levels for the post-Soviet era. Military and security spending at up to 40% of the budget is back to late-Soviet era levels -- though such spending is necessarily at the expense of all other spending for Russia's people. Russia has also been able to keep its oil sales nearly at pre-war levels, at 9.5 million barrels per day, making up for Western embargoes by selling to India and China. The country has evaded sanctions by buying and deploying a "shadow fleet" of hundreds of tankers. One world leader commented privately: "2024 will be much more positive for Putin than we thought. He has managed to reorganize his own industry more efficiently than we thought."

Of course, this form of economic growth has greatly increased Moscow's dependence on oil revenues, on China, and on non-productive war spending. As demand for oil and gas peaks, and competitor production from the Arabian Gulf comes on-stream in the coming year, Russia will be exposed. GDP growth is primarily from war production, but the production of tanks and shells that are then blown up in Ukraine is merely squandering the money for no long-term gain. In the meantime, Russia has experienced a brain drain of some of its most talented citizens.

The Western strategy has not been to destroy Russia's economy, but to tie it down -- restricting access to technology, raising costs, limiting revenues, and make the conflict unsustainable long-term. One US official said: "We'd rather Russia uses its money to buy tankers [for hauling oil] than tanks."

Russia is still likely to persist economically through 2024. The Kremlin is trying to buy time, to wait for the re-election of Donald Trump and reduction of Western funding for Ukraine's defense. That's a faint hope, but it's the only one they have. In the meantime, attention of Western governments is turning back to the central role of those hundreds of billions in frozen Russian financial assets.

Ukrainian President Volodymyr Zelenskyy said: "If the world has $300 billion, why not use it?" Zelenskyy believes those frozen funds should be used to fund Ukraine's rebuilding effort, Britain's Chancellor Jeremy Hunt and Foreign Secretary David Cameron back the move, Lord Cameron saying: "We've frozen these assets. The question is are we going to use them?"

He added that "using some of this money now is, if you like, an advance payment on [Russian] reparations" for its illegal invasion of Ukraine, and could be used "to help Ukraine and save Western taxpayers money at the same time".

The G7's central bankers are considering that move, but it comes with risks. If Russian assets are seized in this way, then what message does it send to other nations -- in the Gulf, Central Asia, or Africa -- about the sanctity of their safe-haven reserves in Western central banks? It's not a simple thing.

Russia's war economy is not sustainable over the long term, and it is heavily dependent on oil exports. Ukraine understands this, and has been ramping up production of long-range kamikaze drones that can hit oil infrastructure deep inside Russia. Efforts to defend those assets further strain the Russian war effort, with coverage being a big problem, and repair of damaged oil facilities is difficult due to sanctions. Russia's Achilles heel is only too obvious, and there's only so much Putin can do to protect it.

* Further investigation with MS ChatGPT counters that Russian government revenues for funding the war are not being primarily obtained through oil exports. According to this view, Putin's primary sources of funds include:

• Generous use of "windfall taxes" on "anything that moves", milking funds from oligarchs and corporations.
  
  
• Seizures of money and property -- particularly of foreign companies divesting themselves of Russia, but less restricted to them as diminishing returns set in.
  
  
• A "National Wellbeing Fund", meaning a sovereign wealth fund set up after sanctions were initially imposed in 2014. It's being drained steadily, but as of the beginning of 2024, there was almost $150 billion USD left.

Putin loots the Russian economy to support his war. Disregarding the specifics, the bottom line is that the war -- economically preposterous and ultimately disastrous; Russia's GDP is less than that of the UK, France, or Germany, making the war unaffordable. However, Putin can't give up -- because he's had it if he does.

[MON 06 MAY 24] THE WEEK THAT WAS 18

DAYLOG MON 29 APR 24: "Legal Eagle" Teri Kanefield, writing in her blog, detailed the weaknesses of the "hush money" case against Trump, saying that Trump was guilty of bad behavior in the case, but didn't necessarily do anything illegal. She then suggested that he was going to be convicted anyway, partly because criminal trials more often than not result in convictions -- but more because Trump's conduct in the trial has been nasty, juvenile, and ignorant, marked by attacks on the judge.

I would have to add that Trump's behavior in the 2016 election was so disreputable that it stands to reason there had to be some identifiable crime in it -- or at least the jury would think so. Of course, Trump will appeal if convicted, but so what else is news?

* In other news, Joe Biden's son Hunter is apparently getting ready to slam a big defamation lawsuit on Fox News. This is gonna be good. It is somewhat surprising that Fox has been spreading vicious lies about people for decades, but only now is being sued for it. It appears that lawyers thought that pressing a defamation suit against Fox would be very difficult and unlikely to pay off. The success of the Dominion Voting Machines lawsuit against Fox appears to have opened the floodgates. I'm waiting for Ray Epps to get his turn.

DAYLOG TUE 30 APR 24: BUSINESS INSIDER had an article on the failure of US weapons in Ukraine, focusing on the "Ground Launched Small Diameter Bomb (GL-SDB)". It was a 112-kilogram (250-pound) glide bomb, normally dropped by aircraft, mounted on an old MLRS booster rocket.

Observers knew that GL-SDB should have been introduced into the Ukraine War early last year, but nothing happened. It wasn't until much later in the year that it became obvious the Orcs had jammed GPS over the battle area, rendering the GPS-guided SDB useless. A US defense official recently publicly admitted that experimental GL-SDB launches in the war zone had failed. The puzzle in this statement was: Why say this now? The trick is that SDB production has presumably moved on to the much more sophisticated "SDB II", which doesn't need GPS.

It seems plausible, if not certain, that the US is now supplying GL-SDB2 in quantity to Ukraine. Some months back there was a report of a Russian assembly in the Donbas region that was hit by 6 long-range munitions and devastated. GL-SDBs come in 6-packs. Maybe an operational test? So the reference to the failure of GL-SDB may be misdirection. Incidentally, there's also a laser-guided version of the original SDB, which could be guided to targets by long-range drones with laser target designators. We'll see in the not-too-distant future.

As a further consideration, the US has developed a finned fuze for standard 155-millimeter shells, the "M1156 Precision Guidance Kit (PGK)", which converts the shell into a GPS-guided munition. GPS, again, is not working in the battle area -- but why not build a version of the M1156 with laser guidance? Or with the ability to home in on "emitters", such as radar or jammer systems? Nice idea, but nobody's saying a word.

DAYLOG WED 01 MAY 24: A UK penpal told me that British media has been calling the US presidential race "neck & neck" -- which means they're listening to US media. I told him that there's very little visible enthusiasm for Trump on the ground here.

It is hard to understand how Trump could conduct a credible presidential campaign from criminal court. I'm still thinking that in 2050, the younger generation will read about Trump and say: "You're lying. This couldn't have happened." Oh, but it did.

I'm not understanding exactly why the polls are so off these days. I don't think there's that much bias, though there may be methodological problems. I suspect most of all that the people being polled are simply muddled. There are some people on the Left who are waffling on Joe Biden, and a lot more people on the Right who have written Trump off, but don't want to admit that voting for him was a bad idea. In any case, Trump's not going anywhere but down.

DAYLOG THU 02 MAY 24: Retired US Army General Wesley Clark, once a NATO commander, recently commented in a public forum on the Ukraine War -- saying it was unlikely the struggle would end one way or the other any time soon.

Clark said: "I don't think Ukraine is willing to compromise and give up its territory. I don't think the Ukrainian military is going to collapse, even though things are pretty tough right now until that American assistance gets there. And I don't think Russia is willing to give up, so it's an endurance contest, and it's also a contest of national will, and whether the United States and NATO will stand firm to support a beleaguered democracy."

Clark said we have to be patient and stay the course: "For Ukraine to collapse, we'd be faced with a huge national security crisis in Europe, and it would empower the Chinese to go after Taiwan. So we can't measure it on the clock."

Clark concluded: "It's going to be a hair-raising time in Ukraine and for the West and then ... summer will probably end with a Ukrainian counteroffensive that regains much of the territory lost. 2025 may be decisive."

I think a counteroffensive is likely, but I'm thinking it will be focused on Crimea, using imaginative tactics. 2025 is indeed likely to be decisive, since Putin will then run out of money to fund the war. If he doesn't win now, he never will.

AND SO ON: This last week, THE ECONOMIST magazine conducted an interview with French President Emmanuel Macron, with Macron saying that Vladimir Putin posed a dire threat to Europe, and that NATO nations should consider direct intervention in the war if necessary:

QUOTE:

END_QUOTE

Macron said:"if Russia decided to go further, we will in any case all have to ask ourselves this question" of sending troops, describing his refusal to rule out such a move as a "strategic wake-up call for my counterparts." Macron described Russia under Putin as "a power of regional destabilization" and "a threat to Europeans' security". He elaborated:

QUOTE:

END_QUOTE

Macron said that Russia "has broken all the frameworks and has basically returned to a logic of total war." However, Putin is overstretched and cannot sustain his war: "Devoting a third of its budget to defense is not sustainable for a country whose gross domestic product is lower than that of France, Germany, or the United Kingdom."

Most NATO nations said they had no intention of sending troops to Ukraine. The Poles, however, are playing their cards close to the vest; if France intervenes, it's a good bet the Poles will, too. Macron was not specific about what form such an intervention would take, preferring to maintain "strategic ambiguity", but insisted that the threat should not and could not be underestimated -- that Europe faced not only a military, security, and economic threat, but a threat to democratic rule.

Macron went on to warn voters to steer clear of nationalists in upcoming European Parliament elections:

QUOTE:

END_QUOTE

He singled out the French Rightist Rassemblement National (RN) Party, saying the RN ...

QUOTE:

END_QUOTE

That's not so true in the USA, with Trump and MAGA disintegrating in full public view. Of course, that's not so easy to see from the other side of the Pond.

[FRI 03 MAY 24] JOE BIDEN (24)

* JOE BIDEN (24): Joe continued his preparation for the Bork hearings, while making campaign stops. He felt he was building up momentum, though not all the media agreed. Then, only days before the Bork hearings began, the reporter Maureen Dowd of THE NEW YORK TIMES published an article in which she accused Joe of "lifting" Kinnock's speech at the Iowa State Fair. Buried in the article, she did concede that Joe had properly cited Kinnock in earlier speeches and that she had contacted the Biden campaign -- which of course called it inadvertent, an oversight. What she did not admit was that she had been tipped off to the gaffe by a staffer of the Dukakis campaign named John Sasso, who helpfully sent supporting materials to the media. Dukakis himself knew nothing about it, and fired Sasso when he found out.

Joe spent much of the weekend before the hearings on the phone with reporters, trying to clarify the situation. Unfortunately it got worse, particularly after the reporters found out about the Bobby Kennedy quote that had been buried in Joe's speech in Sacramento early in the year. When Joe gaveled in the Bork hearings on Tuesday morning, the Right media was in full steam, one essay being titled "Plagiarizing Joe". The Right was already activated in defense of Judge Bork, and Joe had just handed them a mallet to bonk him over the head with. What particularly discouraged Joe was that his integrity, his "word as a Biden", was being so loudly attacked.

In any case, the hearings began on schedule. They started out civil and would remain so, though Joe noticed with some satisfaction that the swing voters on the Judiciary Committee made it clear to Bork they had difficulties with his legal philosophy. When Joe first spoke at length, he assured Bork that the hearings would be fair, expressed his esteem for Bork's reputation as a jurist -- and then asked Bork to clarify his philosophy. Bork did so:

QUOTE:

END_QUOTE

Joe found Bork highly confident, wondering if he might stray over the centerline into arrogance. Joe then brought up specific cases, with Bork generally responding that he had no casual opinion on those cases and would need to work on them to come to any judgement. Joe sensed a weakness and brought up the case of GRISWOLD V CONNECTICUT, which overturned laws against birth-control on the basis of privacy rights. Bork replied that judgement did not establish that there was any such "unstructured, undefined right to privacy" in the Constitution. The Constitution had no soul that could be discerned by the judiciary. Joe moved in:

QUOTE:

END_QUOTE

Bork replied: "I have never engaged in that exercise." Bork, having been given enough rope, had hanged himself: he could say the Constitution had no soul, but most legal scholars and almost all layfolk thought it did. Joe wrote later:

QUOTE:

END_QUOTE

Joe wanted to make sure Bork's assertion was publicly underlined, asking in the following exchange: "... Does the majority have the right to tell a couple that they cannot use birth control?"

Bork effectively repeated his assertion: "... I have never decided that case. If it ever comes before me, I will have to decide it." Bork was saying a couple had no unarguable right to use birth control. [TO BE CONTINUED]

[THU 02 MAY 24] SCIENCE NOTES

* SCIENCE NOTES: As discussed in an article from SCIENCENEWS.org ("Here's How Cool a Star Can Be and Still Achieve Lasting Success" by Ken Croswell, 23 August 2021), the smaller a star is, the more common it is, with our Galaxy heavily populated by little red dwarf stars. There are more, even smaller, brown dwarfs, but they're not generally regarded as true stars, not supporting any more than transient fusion reactions.

It's often hard to distinguish between red and brown dwarfs, because when young they both look the same: red and dim, glowing with the heat of their formation. However, in brown dwarfs fusion reactions in the core quickly sputter out, and they never become true stars. Astrophysicists Dino Hsu and Adam Burgasser at the University of California, San Diego and their colleagues decided to investigate the dividing line between the two.

They determine the breakpoint by examining the orbits of red and brown dwarfs around the center of our Galaxy. When a star is born, it orbits around the Milky Way's center on a fairly circular orbit -- but over time, gravitation tugs giant gas clouds, spiral arms and other stars make their orbits more and more elliptical.

Red dwarfs burn their fusion fuel slowly relative to their size, so they live longer than bigger stars; their predicted lifetimes are far longer than the current age of the universe. However, since brown dwarfs don't, in maturity, perform fusion reactions, those that are still warm are young. That means, on the average, red dwarfs should follow more elliptical orbits around the Galaxy than young brown dwarfs do.

Hsu's team inspected 172 red and brown dwarfs of different spectral signatures, which vary as per their surface temperature. The researchers found that a sharp break in stellar motions separates warmer objects, which on average have more elliptical orbits and are older, from cooler ones, which on average have more circular orbits and are younger. This break appears at a spectral type between L4 and L6, corresponding to a surface temperature of 1,500 to 1,700 degrees Kelvin -- as compared to our Sun's surface temperature of about 5,800K.

Above this critical temperature, the dim suns are a mix of long-lived red dwarfs and young brown dwarfs. Below this temperature, Hsu says: "It's all brown dwarfs." -- though he adds: "We need a more complete sample." That promises to be tricky, since red and brown dwarfs are common, but hard to see. It is still doable.

* As discussed in an article from NATURE.com ("Could AI-Designed Proteins Be Weaponized?" by Ewen Callaway, 8 March 2024), analysis and design of the elaborate structures of proteins is a tricky job, traditionally demanding a lot of computing power -- and even then, not easily getting results. However, machine-learning (AI) systems, through their "educated guesswork", have been able to get good results, opening the door to a new era in "protein engineering".

That new era poses both promises and hazards, hazards including the use of engineered proteins as bioweapons, for example toxins or highly transmissible viruses. Researchers have accordingly now set up a voluntary initiative to promote the safe and ethical use of protein design. David Baker -- a computational biophysicist at the University of Washington in Seattle, who is part of the initiative -- cautions that people shouldn't over-react: "The potential benefits of protein design [by AI] far exceed the dangers at this point."

To assess the potential for malevolent use of designer proteins, Baker's Institute of Protein Design at the University of Washington hosted an AI safety summit in October 2023. Baker said the summit posed the question: "How, if in any way, should protein design be regulated and what, if any, are the dangers?"

The initiative that resulted calls on the biodesign community to police itself, to regularly review the capabilities of AI tools and monitor research practices. Baker wants to set up an expert committee to review software before it is made widely available, and to recommend "guardrails" if necessary.

The initiative also calls for enhanced screening of DNA synthesis, with DNA of course being the tool used to create proteins. At present, many companies providing DNA synthesis services are members of an industry group, the "International Gene Synthesis Consortium (IGSC)", that requires them to screen orders to identify harmful molecules such as toxins or pathogens.

James Diggans -- head of biosecurity at Twist Bioscience, a DNA-synthesis company in South San Francisco, California, and chair of the IGSC -- believes that AI should be used to detect threats: "The best way of defending against AI-generated threats is to have AI models that can detect those threats."

Governments are taking an interest in AI and biosecurity risks as well, though Baker isn't enthusiastic, feeling that government intrusion would interfere with progress in the field. Not all agree. Mark Dybul -- a global health policy specialist at Georgetown University in Washington DC who chaired a 2023 report on AI and biosecurity -- is enthusiastic about the effort, but adds that: "We need government action and rules, and not just voluntary guidance".

David Relman -- a microbiologist at Stanford University in California -- similarly believes that researchers can't be trusted to regulate their own research: "Natural scientists alone cannot represent the interests of the larger public."

* As discussed in a NASA press release "Unlocking Earth's Microscopic Mysteries With NASA's PACE" by Kathryn Hansen, 9 February 2024), on 8 February 2024, the NASA "Plankton, Aerosol, Cloud, ocean Ecosystem (PACE)" satellite was launched from Cape Canaveral on a SpaceX Falcon 9 booster.

PACE's primary payload was the hyperspectral "Ocean Color Instrument (OCI)" designed to observe oceans and other water bodies across a spectrum of ultraviolet, visible, and near-infrared light. OCI will map the distribution of photosynthetic marine microorganisms, or "phytoplankton", to monitor the state of the marine environment. PACE also carried two polarimeter instruments for atmospheric particulate analysis, providing researchers with data on atmospheric aerosols and cloud properties, as well as air quality at local, regional, and global scales.

[WED 01 MAY 24] GEN-AI ARRIVES (12)

* GEN-AI ARRIVES (12): The first wave of Vectara's customers was heavily concentrated in customer support and sales -- areas where a 3% error rate is generally seen as acceptable. It is not acceptable in many other fields. Vectara's Ahmad says he's seen growing interest from the legal and biomedical fields. That error rate may not be too difficult for the legal profession, as long as lawyers realize they have to check their work, but it would mean huge problems for medical service providers.

OpenAI is careful to point out that users should double-check information they get from a GAI chatbot, and for critical issues such as legal, financial, or medical advice, chatbots are no substitute for professionals. AI experts say that LLMs need clear constraints and often expensive product work to be reliable enough for most businesses.

A study performed at Stanford University in early 2024 highlighted the difficulties with GAI chatbots, with the researchers asking ChatGPT basic medical questions to test its usefulness for doctors. To ensure better responses, the researchers prompted ChatGPT with phrases like: "You are a helpful assistant with medical expertise. You are assisting doctors with their questions." Strange as it may seem, experience shows that giving an LLM a little pep talk yields better responses.

Disturbingly, the study found that GPT-3.5 and GPT-4 both tended to give very different answers when asked the same question more than once. Hallucinations were very common, ChatGPT only yielding medically correct answers to queries about 20% of the time. However, the study did conclude that responses from ChatGPT "to real-world questions were largely devoid of overt harm or risk to patients."

Google is among the big AI providers who are now offering products to make LLMs more reliable. Warren Barkley -- senior director of product management for Vertex AI at Google Cloud -- says: "While AI hallucinations can happen, we include features in our products to help our customers mitigate them," Barkley says that Google gives companies the ability to tie -- or "ground" in AI terminology -- LLMs to public data sets, Google search results, or proprietary data.

Not everyone working on GAI systems is all that worried about hallucinations, most notably OpenAI CEO Sam Altman, seeing them as more of a feature, not a bug. Isn't generative AI supposed to be imaginative? Andrej Karpathy -- the former head of AI at Tesla and now at OpenAI -- commented that he struggles a bit with the "hallucination problem" in LMMs, because, in a sense "hallucination is all LLMs do -- they are dream machines."

Amin Ahmad of Vectara believes that hallucinations will be "solved" in a year or two, which Altman sees as plausible as well -- though Ahmad is careful to qualify that statement: "When I say solved, I mean it's going to be that these models will be hallucinating less than a person would be. I don't mean zero."

Even if the current generation of LLMs don't vastly improve, Ahmad believes their impact will still be monumental, in part because we'll just figure out what they're good for:

QUOTE:

END_QUOTE

[ED: For myself, I find GAI chatbots extremely useful, but in part because I never expected them to be magic answer machines. Instead of me scouring the net for information and putting it together, the chatbot scours the net and puts it together itself. If I need clarification or want verification, I can adjust my query accordingly. The chatbot lists the sources it's using, and I can trace back to them if I need to. On simple straightforward questions, the chatbot generally does very well. If it can't give any good answer or a lead, that may suggest I can't find the answer on the internet. Hey, it happens.

Now, I've got myself a subscription to DeviantArt DreamUp, and do have problems with hallucinations there -- mangled hands, three legs, floating fingers, strange eyes, and so on. I think that will get better with more training of the engine. Oddly, the hallucinations seem to get worse if try to make too detailed a specification; it seems to confuse the engine. Give a short simple prompt, the engine will follow its own instincts and do a good job, though the result may not be what I want.]