2025-06-23 16:53:16

the Trakaido project continues.

Hopefully today (or possibly tomorrow) it will be at a "release milestone".


The three main issues I am dealing with:

  • LLMs are bad at state machines, and React has a lot of them.
  • LLMs are bad at succinct UIs, and mobile apps need them.
  • The app was architected so most of the code was in Trakaido.jsx, and passed to the "Modes". But, with additional modes, it is now three-layers (Trakaido to Mode to Activity). And the functions being passed (like playAudio) should be imported directly by the activity. This refactor, somehow, is beyond the LLMs unaided capabilities.

So, after 90 minutes of trying and failing to get it to work, I reverted and am doing a more hands-on approach of re-factoring.

2025-06-22 02:37:52

Seen on Substack ⚙️ https://substack.com/@tedgioia/note/c-128027920 :

Yesterday I did a Google search to identify the most successful movie of the last 5 years. Instead I got an AI response—which I didn’t ask for. AI identified a film from 2019 as the most popular movie in the last 5 years. And it even specified that the movie came out in 2019.

AI always serves up this garbage—it makes mistakes a child would easily avoid. And now tech CEOs want AI to rewrite history? It won’t even get the dates correct.

It gets worse, Google AI seems to think that the period from 2019 through 2024 is five years in duration. It can’t even count to five without making a mistake.

The period from 2019 to 2024 IS five years. Yes, the whole period would be six years. But if you want a period of 5 years that ends in May 2024, it is 2019-2024.

That Samaritan's entire schtick seems to be saying that AIs are garbage because they didn't make the same mistake that he made. And, that the solution is for more people to get an Oxford education like he did.

⚙️ https://www.honest-broker.com/p/5-ways-to-stop-ai-cheating

🔥 It is both a humblebrag and a completely stupid idea. "Use more Latin! Don't even allow typewriters!" His solution is to stick one's hand in the sands-of-the-past, and assume that this will fix everything. Because he's not a technologist, he's an old musician who thinks he knows everything but very clearly does not.

fuld v. plo US Constitution
2025-06-20 15:54:03

Fuld v. PLO ⚙️ https://www.supremecourt.gov/opinions/24pdf/24-20_f2bh.pdf is out. While I don't like the decision, it is hard to make an argument that they could have ruled any other way.

The PSJVTA’s personal jurisdiction provision does not violate the Fifth Amendment’s Due Process Clause because the statute reasonably ties the assertion of jurisdiction over the PLO and PA to conduct involving the United States and implicating sensitive foreign policy matters within the prerogative of the political branches.

The primary reason why the statute should fall is that the United States inherently doesn't have jurisdiction over a quasi-state entity halfway around the world. But, this isn't based in a concept of "due process". If anything, by asserting a right to due-process, the PLO does implicitly consent to jurisdiction, in a way that "not ceasing a policy" does not.

The "foreign policy" note is more concerning. There is a 🔥 far-right theory of government that would hold that the Constitution only governs how the federal government interacts with US citizens, and does not bind its external actions 💡 other than a few enumerated exceptions, such as "participating in the slave trade". I disagree with this; and I generally feel that the blanket exception various courts are working towards is a loophole large enough to drive a truck through.

But, the limits of American power will remain evident. A court can issue as many universal injunctions as it wants, but the PLO will not act based on it. And the ever-increasing fines based on foreign activity by a kangaroo-court will impugn the United States more than they will ever punish the PLO. When Russia fines Google an amount so large the TV announcer cannot pronounce it ⚙️ https://www.bbc.com/news/articles/cdxvnwkl5kgo , it is dismissed as the folly of a rogue state. 💡 Clarence Thomas, in his concurrence, spells it out more thoroughly. That Congress may override general principles of international law does not imply that it should, but instead that the relevant considerations are not constitutional ones. If you view international law as superseding the constitution on certain matters, it should not be surprising that the Constitution does not incorporate these restrictions.

2025-06-19 03:46:34

Another round of desired feature requests.

This set is primarily "gameplay" focused, with getting the "audio" right a primary goal. The secondary goal is getting the "tools to generate wordlists" working. ⚙️ those are "greenland" projects, where the language-corpus and LLM queries for translations live

Click title to read full message...

2025-06-19 01:55:08

https://blog.brianbalfour.com/p/the-next-great-distribution-shift 🔥 my verdict: it stinks!

There are several glaring factual or interpretation errors that cloud his analysis of the space.


But first, his analysis is that "tech platforms" go through a three-stage cycle:

  • Identify the moat
  • Open the gates
  • Close for monetization

This is, on some level, accurate. There are a lot of people trying this on Substack, or elsewhere. Give it away to get viral growth when you are small, charge and profit when you are big. Business 101.

His analysis of how this worked for multiple large tech companies, unfortunately, is fatally flawed.


Facebook - to start with, several of the "added features" like Marketplace and Photos existed years before the ill-fated "Facebook Platform" ever launched. But, the real problem is that Zynga-style gamification cluttering activity feeds was a disaster 💡 although, in hindsight, it was probably better than the political takes that followed it. They could not allow it forever, so it stopped. And, without a limitless reservoir of free advertising, Zynga crumbled.

Also, constantly throughout this post, the author is nostalgic for what can best be described as awful crap. The "quizzes" and "vampire bite" apps were not something to be defended; the people who ostensibly "made millions" from them not to be celebrated.


Apple - first off, the "70-30" revenue distribution is considered as both "open" and "closed". Once again, people being able to make millions off the "iFart" app is not to be celebrated. And the demands that the platform never include features that are in platform apps are unreasonable.

But the more serious concern is the specter of government regulation. Most of the "privacy restrictions" he criticizes are imposed by the government. And the anti-trust concerns govern pricing more than anything else.


Google - the SEO ⚙️ search engine optimization industry is largely the scourge of the earth, and should be destroyed. 💡 there seems to be a trend that it is bottom-feeders, creators of frivolities, and abusers of the public common that suffer from this "contraction". Unlike the author, overall I would prefer those changes

The problem here is that the moat is contrived and unrelated to the changes. The increase in oneboxes ⚙️ in Google parlance, a "onebox" is a structured-content response, other than organic search or an ad. For example, a "weather" onebox for a search for the weather. They were called "oneboxes" because there would be only one per search result page, but that seems to have gone by the wayside. was certainly a change from the 2004 philosophy ... but it's unrelated to market-share growth against Yahoo.


LinkedIn - I'm not even sure what he's talking about. I don't believe there was a "2 year window" where LinkedIn encouraged content creators. I think it was either a COVID effect, or a personal observation of this guy extrapolated to the whole site.

And, once again, the people creating "B2B Marketing" content are people I do not want to succeed.

as a reminder, viruses are bad. these people act like cockroaches, and then are offended that people want to make them go away.


So the conclusion, that platform cycles are accelerating, cannot be supported by this data. And the attempt to project this onto ChatGPT is bad.

The moat has to be chat memory ... because this guy can't think of anything else. And, even if it is ... the platform that can access your email history is more valuable than the chat history.

Also, ChatGPT can't shut down the API. For one, it's profitable. For two, the alternative of "self-hosted models" is too good for this to meaningless harm the (now)-competitors.

The ChatGPT "market share" advantage is driven by higher name recognition among non-technical people, and people who haven't tried competitors. They are in the position of Yahoo! and MySpace. I'm not saying that this means they can't be successful ... but there is not such an advantage that Google and Facebook are irrelevant.

And, finally, the complaints that ChatGPT is competing with enterprise players like Glean are pathetic. Presumably the feature is "connectors" ⚙️ https://techcrunch.com/2025/03/17/openai-to-start-testing-chatgpt-connectors-for-google-drive-and-slack/ : being able to access data in your Google Drive account. This is a feature that Claude has had for a long time. Once again, this is a ludicrous demand that a platform remain incomplete because someone else implemented a necessary feature.

2025-06-18 22:46:42

The market for learning the Lithuanian language is quite small.


While the market for "learn Spanish as a second language" is bigger, the much more valuable tool is one that teaches English as a first language.

What does it need?

Click title to read full message...

2025-06-18 18:28:46

🔥 I only assume these people are being indirectly paid by the fossil-fuel industry. In general, for these bad-faith, low-quality arguments against any form of renewable energy, it is a safe bet. Regardless of whether they are paid, directly or indirectly, it is very blatantly propaganda.

⚙️ "Center of the American Experiment’s mission is to build a culture of prosperity for Minnesota and the nation. Our daily pursuit is a free and thriving Minnesota whose cultural and intellectual center of gravity is grounded in free enterprise, limited government, individual freedom, and other time-tested American virtues."

https://jasonhayes.substack.com/p/net-zero-just-isnt-happening

LINKS TO

https://files.americanexperiment.org/wp-content/uploads/2025/06/Shattered-Green-Dreams.pdf

There are so many flaws in this report, it is enough that one should throw the report (and its advocates) out entirely. If a person gives you 10 arguments, and 8 of them are objectively wrong, and you aren't sure about two, you must not give that person the benefit of the doubt.


ONE. Bad Math.

There are a remarkable number of situations where they give highlight, where they highlight figures such as: "This resource will consume 10x production" or "2000%". They never emphasize the baseline. For example, when they talk about nickel production, they say it is "1600%" of American production but "11%" of global production. Similarly, for aluminum, they say "45 times the U.S.’ 670,000 metric tons of primary aluminum production in 2024 and 42 percent of the world’s 72 million metric tons".

Yes, if America doesn't produce much nickel or aluminum you will get a very large multiple of that. The solution is either recognizing that it's a global economy and that an America-first-at-all-costs approach is simply a smoke screen to shut down something you don't like, or building more nickel mines in the US; a large percentage increase that might not be a large absolute increase.


TWO. Environmentalists whenever it helps their cause.

The piece is very aggressive on highlighting any actual or perceived environmental ills that can be traced to solar or wind energy. 200 turtles don't have an habitat? Crime. Use a dirty mine in Indonesia? Crime. However, for systems they support, these types of environmental ills are simply ignored or dismissed.

Furthermore, many of the scenes they highlight are purely speculative or are described as topics that may need additional research. "Do wind turbine noise cause whale problems? There is no evidence of it, but they aren't sure. We can't take the risk." This once again is caution at all costs because they have an underlying motivation to kill the project.

And when there is some merit to their claims, they are still exaggerated, hyper-aggressive and concerned with relative impact only when it benefits them. They claim a study says 350,000 birds died in wind turbines in one year. They then say this is definitely an underestimate. With no argument other than that it helps their cause. Meanwhile, with a U.S. population of 15 billion birds, most of which have a lifespan of 5 to 10 years, this seems like a relatively low number. They note in passing that "building collisions" caused 600 million deaths, but shrug it off, saying that this high number means "any cause of mortality needs to be tackled".

🔥 If your philosophy is that one dead bird is too many, nothing will satisfy you. I really don't care to rebut those arguments further.


THREE. Focusing on the failures.

The report talks at length about the Ivanpah Solar Power Facility in California. This is not a traditional solar panel plant, but a plant that uses large arrays of mirrors to superheat a reservoir to over 1000 degrees Fahrenheit. This design has several flaws and does cause more serious safety issues. The plant is also being decommissioned early because it is ineffective. However the authors extrapolate from this to say that all solar plants have these same problems. This is a sleight of hand deception.


FOUR. Focusing on meaningless metrics.

Like several other influencers in this space, they focus heavily on one metric that suits their goals, the energy generated per land area used. A solar plant that uses 70 square miles and a nuclear plant that uses one square mile may generate the same amount of energy.

This is problematic for several reasons. First, the land used is not equivalent. 70 square miles of the Nevada desert may not be used for anything else. Thus, the cost is low. Second, if a solar plant replaces a field growing ethanol corn in Iowa, it is done because the economic and the energy returns are far greater. Simply claiming that some other technology 💡 which has its own drawbacks might be even better doesn't mean you can't do the first improvement.

A very different meaningless metric comes from long-time anti-solar shill Robert Bryce, who is quoted as saying that in Minnesota, “nearly all of the wind projects” are “located in counties that are poorer than the statewide average”. Except of course they are. Once the 9 counties of the Minneapolis-St. Paul metro area (the 9 richest counties in the state) are excluded, only 3 of the remaining 78 counties are not "poorer than the statewide average". This is not a legitimate socio-economic argument, it is deception.

And a third one comes from another piece linked in the blog post ⚙️ https://lifepowered.org/data-banning-fossil-fuels-wouldnt-stop-climate-change/ where they claim that "if the United States banned fossil fuels by 2050", it would only cause "a tenth of a degree" of change by 2100. The (hidden) assumption is that no other countries make any changes. The argument is effectively saying "if we divide the world into 20 parts, none of them can make a big enough change on their own, so we should do nothing". 🔥 Which is a frightfully stupid reason to do nothing.


FIVE. Misunderstanding system lifecycle.

They have one quote which massively misrepresents the lifecycle of these products. They say that most solar panels only have warranties of 25 years, whereas a coal plant can be in operation for 60 years. 💡 The concerns about the lifespan of wind turbines seem to be at least somewhat based in fact rather than misrepresentation. The difference is that the word warranty is used.

Nobody can offer a 60-year warranty on solar. I looked into it. It simply doesn't make sense. The problems related to the fact that neither the purchaser, the manufacturer, nor the insurer are likely to be around in 60 years, either as corporate entities or natural persons, make such a long warranty become meaningless.

This doesn't mean that technology can't last that long. It simply means that the financial product of a warranty becomes meaningless. For the coal plant, there is no warranty. The owner of the plant is incentivized to keep it running as long as it is economical.


SIX. Obstacles of our own design.

The authors consistently point to tax credits and other government incentives that encourage fast depreciation and replacement of solar or wind, claiming that this proves they are not sustainable in the long run. Similarly, they point to the permitting process, at one point making the ludicrous claim that it would take 1,400 years to prove all the permits necessary for a net-zero energy transition.

The flaw here is obvious: These are stupid rules the government made up. They can just get rid of the rules. You don't want them to get rid of the rules because you don't want the solar plants. This isn't a real obstacle; it's a fake one.


SEVEN. Failure to consider alternate technology.

They routinely math based on an assumption that batteries need to support 100% of energy storage needs. This is a fallacious assumption. In particular, for cold weather climates, a battery-based system will never be feasible. The only possible option is a solar and/or wind-powered system that generates some amount of (presumably) methane that can be stored and then burned during the winter season. This would have a substantially lower metal, et cetera cost. However, it doesn't fit their conclusion. So they need not consider it.

2025-06-18 17:04:31

https://alluvial.substack.com/p/finding-value-in-faraway-places

I have a few thoughts on this piece, none of which lead towards anyone in the U.S. should be getting involved in this.

  • There's a certain sense that investing in a company in Papua New Guinea that is generating massive returns on capital feels immoral. Where is this money coming from, and why are you taking it? Isn't this just a new form of colonialism?
  • How much of this is, the stock market isn't real? Or, the stock market in the US actually is real but this one isn't. There are no "bid/ask" prices for many of these. There is no "price discovery". One company can trade at twice its value, another at half its value.
solemnization Miscellany
2025-06-13 18:52:03

https://x.com/patio11/status/1930340924212539902 (n.b. This is extremely well-known among companies which have a business process where you sign things. Most of them use a signature to demonstrate solemnization rather than authorization or authentication.)

I'm not familiar with the word solemnization here. But the concept makes sense. The purpose of the signature isn't to demonstrate consent, or identity. It is to create a ritual.

2025-06-13 18:14:52

An early version is available at https://earlyversion.com/trakaido .

Still to be done:

  • The "Journey" mode has several issues. The logic of gradually rolling out new words 🔥 and, of course, SPACED REPETITION!!1! isn't done yet.
  • The layout on mobile devices is somewhere between mediocre and unusable.
  • The audio files are only about 98% good, rather than 100%. ⚙️ the OpenAI API is not suitable for reviewing audio files. I don't want to spend a week building an "Audio Lab" that will more deterministically assess is there breathing in the file or are the sounds all correct. Or do I?
  • There are a few bugs where the audio doesn't play when/where it is supposed to.

The list of things that were done, beyond the fold:

Click title to read full message...