Saturday, August 05, 2023

Pt. Molate

 









The Big Screen

 One recent midweek morning, people wearing various combinations of white, orange and black could be seen maneuvering their way from various points in the Bay Area to downtown San Francisco. They came on foot or by cycle, bus, train or cable car, in a Lyft, Uber, or private car or maybe a boat or even a scooter.

Many of them carried baseball gloves, most wore sunglasses (mostly pushed up for now), some carried backpacks, blankets, water bottles or cups of coffee or stronger beverages cleverly disguised, or so they thought.

It was a typical San Francisco summer’s day — a thick layer of fog blanketed the entire area and a soft breeze blew in from the west, where the Pacific Ocean sloshed ashore with its familiar rhythms. Somewhere a foghorn moaned.

In what is commonly known as McCovey Cove, just off right field at the San Francisco Giants’ baseball stadium, the usual array of kayaks, sailboats and other water craft were gathered, while inside the park vendors were preparing their daily stocks of hotdogs, popcorn, crab cakes, garlic fries, Ghirardelli hot fudge ice cream cups, and plenty of beer.

By the time the first pitch was thrown (12:45 p.m.), some 33,000 people had assembled and the home team had taken the field as their opponents, the Arizona Diamondbacks, prepared to hit.

Oracle Park, as it is currently called, is an intimate tiered venue tucked into a nook of the shoreline, with a variety of kitschy features beyond the outfield wall, including a giant replica of an old-fashioned baseball glove and an equally gaudy oversized Coke bottle.

Why are they so large? Well the home team is called the Giants, so big stuff is to be expected. But the main feature, the one that draws everyone’s attention, is a massive scoreboard that includes a huge video screen that captures virtually every aspect of the game for instant replay.

All eyes go out to the screen repeatedly as the crowd tries to figure out what just happened in real time. “Was that runner safe or out? Let’s watch the replay.” 

While there are still many in the seats who are current or former players and coaches of all ages, and therefore intimately familiar with the intricate mechanics of the game, others are just here for a good time, for the beer and hot dogs, and the relatively remote possibility that they might catch a foul ball. (Thus the gloves.)

On this particular day, it was a pitcher’s duel, and the home team ultimately prevailed 1-0, thanks to a solo home run hit out toward McCovey Cove by Giants first-baseman LaMonte Wade, Jr., a stellar pitching performance by a bevy of pitchers plus several very good defensive plays.

The visitors played well too, but just not well enough. One striking feature of the modern game is that all the players and umpires look up at the giant screen after each play just like the fans. It’s as if no one is entirely certain about what sctually happened until they see it on TV.

Baseball is a game of tradition — a century and a half of it — but it is also changing with modern technology.

Nowadays there is a “challenge” feature, whereby managers can demand a review of a close call. When this happens, all the camera footage is transmitted back to a set of umpires in New York and everybody in the park just sits and awaits their verdict. 

During this interregnum, tempers in the dugouts and on the field have a chance to recede and as a result very few coaches or players fight or get thrown out of games any longer — outbursts that were part of the standard fare in years past.

And so onward it goes. After the game, the boisterous home crowd happily packs up its gear, which on this day includes a promotional pack of trading cards, to stream out of the brick-walled stadium back onto the sidewalks, bikes, motorcycles, buses, trains, cable cars, Lyfts, Ubers, private autos, boats and scooters to head back homeward.

By now it is mid-afternoon. The sun is starting to break through the fog above San Francisco Bay, and it’s finally time to push those sunglasses down into place. The winds are kicking up, too; in response the locals just add another layer of orange, black or white.

Tony Bennett’s voice comes on the loudspeaker. It was a very good day for the home team.

LINKS:

Friday, August 04, 2023

Deal or No Deal?

It seems fair to say that so far as the American system of democracy is based on the rule of law, that system is under fierce attack in the matter of the Trump indictments.

In fact, the situation is unprecedented and may easily devolve into a terrible crisis.

The country is split roughly in two. One side believes he must be held accountable for inciting an insurrection and attempting to steal the 2020 election.

The other side believes his unsupported assertions that the election was stolen from him and that he is the victim of political persecution.

Given the messy way both democracy and the criminal justice system actually work, it may turn out that neither side will ever gain complete satisfaction.

The court cases now under way are occurring under the cloud of the 2024 election cycle. They’ve taken prosecutors years to prepare and now time is growing short for them to conclude before reaching the point they truly would interfere with those elections.

That would be an unacceptable breach of the norms that help preserve the legitimacy of our democracy.

As for those who buy Trump’s argument that he the victim here, they eventually are going to have to accept the truth that he is his own worst enemy when it comes to what got him into legal trouble. This all could have been avoided but for his actions as a sore loser.

I can’t help thinking that this entire messy chapter in our history could be solved at least in the short term by a deal. Trump pleads guilty to some charge, and is barred permanently from ever again holding public office.

That’s pipe dream, I know, but wouldn’t it be nice? The whole psychodrama could go away and we might be able to getting down to addressing the real problems facing us collectively, like climate change, wealth disparity, the rise of authoritarianism, and the existential threats that face our children and grandchildren

LINKS:

 

Thursday, August 03, 2023

State's Right

Once the Georgia indictment of Trump comes down, which will be sometime this month, the full constellation of criminal charges against the former President will be known — two federal and two state prosecutions.

Presumably the federal charges will take precedence in terms of timing, for a variety of reasons.

The two federal cases really need to be tried before next year’s election. Trump is a candidate and voters need to know whether the juries find him innocent or guilty. If he should somehow succeed in getting elected President again, he could easily dismiss the federal cases against him or, if convicted, pardon himself.

In fact the is exactly what he would do. He’s made that clear.

But the state charges are another matter. Both also involve election interference — in 2016 in the New York case and 2020 in Georgia. Even if he were President, he could not easily escape the state cases, nor could he pardon himself if convicted.

Therefore, under the virtually unthinkable scenario that Trump should once again come to power, the only way of holding him accountable for his criminality may be those two state cases.

Thus two prosecutors, one in the North and one in the South, may hold the key to the good guys winning a second Civil War, should it come right down to it. There is an element of poetry to that. 

LINKS:

  • Trump’s Case Has Broad Implications for American Democracy (NYT)

  • The Indictment of Donald Trump Could Save Democracy (Newsweek)

  • The New Trump Indictment and the Reckoning Ahead (New Yorker)

  • Barr calls Trump indictment ‘tip of the iceberg’ in Jack Smith’s case (The Hill)

  • What Makes Jack Smith’s New Trump Indictment So Smart (NYT)

  • Rudy Giuliani Flips Out Over Trump's Jan. 6 Indictment (HuffPost)

  • Judge Tanya Chutkan is a tough Trump critic, toughest Jan. 6 sentencer (WP)

  • DeSantis-controlled Disney World district gets rid of all diversity, equity and inclusion programs and staffers (NBC)

  • Fitch downgrades US credit rating, citing mounting debt and political divisions (AP)

  • Is This the Beginning of the End of the Writers Strike? (Vanity Fair)

  • An end in sight for Hollywood's writers strike? Sides to meet for the first time in 3 months (AP)

  • U.S. preparing order to evacuate U.S. embassy personnel from Niger (Politico)

  • Why the nightmare in Niger is the world’s problem (Economist)

  • Russian drone strikes hit a Ukrainian port on Romania’s border that is key to grain exports (AP)

  • Elon Musk's controversial AI plan has artists jumping ship from Twitter (Creative Blog)

  • Where Thousands Of Tech Workers Went After Mass Layoffs (CNBC)

  • Hackers Keep Finding New and Sophisticated Ways to Use AI for Crime (Decrypt)

  • Automated, AI-equipped security tools help hold the line on data breaches (AI in Healthcare)

  • FBI Warns About China Theft of US AI Technology (VoA)

  • The generative A.I. battle between companies and hackers is starting (CNBC)

  • Movie extras worry they'll be replaced by AI. Hollywood is already doing body scans (NPR)

  • Researchers figure out how to make AI misbehave, serve up prohibited content (Ars Technica)

  • AI improves breast cancer detection rate by 20 percent (Politico)

  • How to detect fake news with natural language processing (Cointelegraph)

  • Kids Explain How Disney Turned Them Gay (The Onion)

 

Wednesday, August 02, 2023

The Test

Donald Trump’s attempts to overthrow our democracy will finally land him in court thanks to Tuesday’s indictment by the U.S. Justice Department. He faces four counts of breaking the law while trying to alter the outcome of the 2020 election.

The 45-page indictment identifies six co-conspiracists, though not by name and at this point they have not yet been indicted themselves. There is certain to be more on them forthcoming.

This sets up a climactic moment for democracy, here and around the world.

The trial has to conclude before the 2024 election. But even if that can happen, it could have a negative impact on next year’s election. My deep worry is that even a conviction of Trump may not be able to prevent the Constitutional crisis I see brewing in the near future.

In this case, the January 6th riot may have been only a prelude. The anger and resentment behind Trump is a sickness that imperils our democracy.

LINKS:

 

Tuesday, August 01, 2023

The Next Word

 If the artificial intelligence craze feels strangely familiar, it should, because we’ve been dealing versions of it for a long time now under different names. We’ve been introduced to products that supposedly get smarter the more we use them.

Our entire world has evolved into a giant software testing environment and we are the guinea pigs.

Meanwhile, there are unique aspects to what is now called AI, and they’re worth knowing about. One of the best explainers to date is in Ars Technica. Here is my summary in the form of excerpts:

  • To understand how language models work, you first need to understand how they represent words. Humans represent English words with a sequence of letters, like C-A-T for "cat." Language models use a long list of numbers called a "word vector." The full vector for cat is 300 numbers long.

  • Each word vector represents a point in an imaginary “word space,” and words with more similar meanings are placed closer together. For example, the words closest to cat in vector space include dog, kitten, and pet.

  • Word vectors are a useful building block for language models because they encode subtle but important information about the relationships between words. If a language model learns something about a cat (for example, it sometimes goes to the vet), the same thing is likely to be true of a kitten or a dog.

  • (But) words often have multiple meanings. And meaning depends on context. To transform word vectors into word predictions, large language models (LLMs) use layers that act as transformers. Each layer adds information to help clarify the meaning of that word and better predict which word might come next. 

  • Researchers don’t understand exactly how LLMs keep track of this information, but logically speaking, the model must be doing it by modifying the hidden state vectors as they get passed from one layer to the next. It helps that in modern LLMs, these vectors are extremely large. The most powerful version of GPT-3, for example, has 96 layers and uses word vectors with 12,288 dimensions—that is, each word is represented by a list of 12,288 numbers!

  • A key innovation of LLMs is that they don’t need explicitly labeled data. Instead, they learn by trying to predict the next word in ordinary passages of text. Almost any written material—from Wikipedia pages to news articles to computer code—is suitable for training these models.

  • You might find it surprising that the training process works as well as it does. ChatGPT can perform all sorts of complex tasks—composing essays, drawing analogies, and even writing computer code. So how does such a simple learning mechanism produce such a powerful model?

  • One reason is scale. It’s hard to overstate the sheer number of examples that a model like GPT-3 sees. GPT-3 was trained on a corpus of approximately 500 billion words. For comparison, a typical human child encounters roughly 100 million words by age 10.

  • At the moment, we don’t have any real insight into how LLMs accomplish feats like this. Some people argue that such examples demonstrate that the models are starting to truly understand the meanings of the words in their training set. Others insist that language models are “stochastic parrots” that merely repeat increasingly complex word sequences without truly understanding them.

  • Traditionally, a major challenge for building language models was figuring out the most useful way of representing different words—especially because the meanings of many words depend heavily on context. The next-word prediction approach allows researchers to sidestep this thorny theoretical puzzle by turning it into an empirical problem. It turns out that if we provide enough data and computing power, language models end up learning a lot about how human language works simply by figuring out how to best predict the next word. The downside is that we wind up with systems whose inner workings we don’t fully understand.

I recommend that you bookmark the entire Ars Technica article.

LINKS:

Monday, July 31, 2023

Corruption, Taliban-Style (Afghan Report 65)

This is another in a series of secret reports from a friend inside Afghanistan.

Dear David:

Transparency International is the NGO that maintains something called the Corruption Perception Index (CPI), and since the Taliban came to power in Afghanistan our country has shown a  marginal improvement in the rankings. The Taliban takes credit for this improvement as it struggles to gain legitimacy in the eyes of Western leaders.

But the supposed improvement is a sham.

For example, a man I know named Husain has worked more than 10 years in government. He was happy having a job that paid most of his family's expenses, but a few months ago, he had to leave his job because of the constant corruption and injustices he witnessed.

He worked as a HR manager. When a seminar or workshop was held by international organizations, it usually included a financial bonus for his department. Even though Husain did the major work organizing the seminar, his supervisor gave the bonus to one of his friends or relatives instead of Hussain. 

In addition, whenever there was a vacancy in their office, the supervisor appointed his friends and relatives or someone who paid him a bribe. Finally, Husain got fed up and quit. Now he is struggling to make ends meet but at least he has his pride.

This type of petty corruption starts at the top of the Taliban government. A man named Gul Ahmad is a friend of the Deputy Minister of Interior. In a letter to the head of procurement, the minister writes that Gul Ahmad should be provided with a car. Gul Ahmad then gets an $80,000 armored car. Friends take care of friends.

So yes, Afghanistan may have improved its CPI ranking from 174th in the world to 150th under Taliban rule, but we here on the ground know that the government is thoroughly corrupt and rotting like a fish from its head. 

Sunday, July 30, 2023

Alone Together

(This is an excerpt from an essay I wrote at the height of the pandemic in late July 2020. At the time my essays appeared daily on Facebook.)

(self portrait)

Loneliness is one of the fellow travelers accompanying the virus as it sloshes over the land. It already was a big problem in American culture, of course, but 2020 is angling to be the loneliest year on record.

“During the beginning of the pandemic, when social distancing and other restrictions were put into place, we found that depression and loneliness were being experienced at considerably heightened rates in the U.S.,” said Molly Rosenberg, lead author of a study in June at Indiana University. “While these restrictions were and continue to be critically important to protecting Americans from the virus, it is clear that the spread and response to the virus has had a tremendous mental health impact on Americans.”

"Many of us feel lonely from time to time and these short-term feelings shouldn’t harm our mental health," says the Mental Health Foundation. "However, the longer the pandemic goes on for, the more these feelings become long-term. Long-term loneliness is associated with an increased risk of certain mental health problems, including depression, anxiety and increased stress.  The impact of long-term loneliness on mental health can be very hard to manage."  

As the months of isolation drag on, some have found ways to cope, but many others have not. It's my impression that social media like Facebook have never been more important in our lives than they are now. Every night, as I write a new post, I hope to be helping relieve some of the isolation all of us feel.

That at least is my intent. If we have to spend so much time alone, maybe we can somehow be alone together.

LINKS (7.30.23):