cloudbonsai

ColdFusion has been maintained by Adobe since 2005. Apparently, they think there is future for it:

    ColdFusion Next is evolving into an AI-powered, cloud-native platform designed for modern business automation.

See “Roadmap for Adobe ColdFusion 2026” for more details: https://www.adobe.com/cc-shared/assets/pdf/products/coldfusi...

For what it’s worth, this article cites a longitudinal study done by German researchers.

They basically installed a bunch of bird feeders outside nursing homes to see if watching birds improves the health of the elderly people. The researchers concluded that it did:

    Nursing home staff were the first to see the difference the birds made. "For weeks, they haven't seen a smile on [a resident's] face, and then they see the smile," Kals says.
    Some of the memory and concentration improvements are evident after just three to four weeks, Kals says.

It seems that they used the following bird feeder in their study:

https://www.lbv.de/umweltbildung/fuer-seniorenheime/ausstatt...

I'm curious about your perspective on the impact of LLMs on physics.

My impression is that (theoretical?) physics is increasingly becoming a branch of abstract math. It has somewhat reached a practical limit for experimental verification -- perhaps the last landmark result was the confirmation of the Higgs boson 15 years ago?

So I see no reason LLMs cannot crank out a lot of theoretical physics papers, maybe combining forgotten/obscure theoretical branches explored in the past. After all, the most advanced theoretical physics is about internal consistency just as pure math is.

It seems that Meta assigns a unique hostname to each shared conversation. For example:

> fiscal-review-b4a5ff.cf.metaaiusercontent.com

So when someone opens the link, the DNS resolution exposes the unique hostname. The original bug reporter found dozens of these URLs in passive DNS logs.

I'm curious why such a large gap exists between our numbers. My hypotheses are:

A. The average utilization rate of inference GPU cards is not that high (i.e. it peaks during the daytime, but is idle at night). So the time window for earning $38/h is fairly limited.

B. They are renting costly GPU capacity from 3rdparty hyperscalers because they can't meet the peak demand with their own hardware.

C. Free-tier users are eating the gross margin.

D. Subscription users are eating the gross margin.

Another observable point is that DeepSeek's company-wide gross margin was reportedly 44.6%. It's intriguing what the full picture looks like here.

Your estimation feels too high. It implies >1000% ROI on a GPU card (= $300k gross annual profit on a $30k card).

As a reference point, Z.ai is operating at 26% gross margin as of 1H26. So assuming ~90% gross margin is likely a bit off.

Zebrafish have retinas very similar to mammals. For whatever reasons, they are able to regenerate their retinas when injured, while we mammals can't.

Scientists hypothesized that the presence of a certain protein is the reason why we can't, and they experimented with mice to see if blocking the protein would allow them to regenerate their retinas. It did.

What’s more interesting to me is the fact that these two kernels (con-leche and con-ron) were actually implemented by Claude (with “human supervision”).

Programming as a hand-crafted art is really diminishing fast.

They created an implementation of a Lean kernel in Lean (con-leche) and verified it as consistent.

Since there can be bugs in its compiler and runtime, they translated the code into Rust (con-ron), and verified the resulting kernel with another Rust-based verification tool.

There is also an ongoing verification project of Lean4, which is written in Lean4 itself. (My mental model is Lean4 := Raku and Lean := Perl)

So we have two verified kernels now, and another is awaiting release.

Yeah. I experimented with a random Japanese news article. Scanning it with my eyeballs, I found:

    Correct .. 80 chars
    Missing .. 12 chars
    Wrong ....  8 chars

So I guess the accuracy is roughly 80%. Maybe useful, but not super reliable.

Reading your comment, I’m curious if AI models will eventually “fork” mathematics. (Sorry for beating the same cranky analogy)

Like, not expecting human experts to review and merge their results, but developing math branches on their own. I guess it’s at least doable for applied math. For example, if AI agents find a fast method of prime factorization, they can demonstrate it by cracking RSA encryption. No human review required.

I’m less sure how it plays out for pure math, though. My intuition is fairly limited here.

Do you think it’s possible to operate math research in a fully autonomous way (i.e. with no human expert involvement)?

As a software engineer, my mental model so far is that OpenAI made a gigantic (but potentially sound) pull request to the math community, and the mathematicians just did not have enough bandwidth to review and merge it.

If my model is correct, I guess it makes a lot of sense to fund mathematicians, because their lack of resource is the bottleneck here.

It’s simple: Because Jim Simons (who founded RenTech and Simons Foundation) was a prominent mathematician himself.

He was also a major backer of autism research, because his family member had it. People spend money to where their passion is.

I’m curious: why isn’t a “second opinion” included in the standard medical protocol?

As an engineer, embedding such a check step (basically a PR review for your body!) in the standard workflow feels very natural. But in the medical context, getting a second opinion usually needs extra effort by the patient and often involves awkwardness.

It feels involving too much friction given the high stakes. What is your opinion on this?

What jonas21 is stating is correct. If you have the absolute best performance, you are automatically on the Pareto frontier because you’re the most cost effective one (or the only model) on that performance tier.

It seems to me that the biggest winners of the financial sector automation, circa 1990-2010, were (1) a small group of sophisticated prop trading firms and (2) the general public, while the biggest losers were traditional bank traders whose jobs were substituted by computers.

I’m not sure how it turns out for the software engineering space this time, though. Will ai-assisted boutique firms outcompete Google (search), Microsoft (OS and office suite), Amazon (e-commerce and infla) or Apple (mobile device)?

My impression is that the top 0.2% are spending a lot on LLM subscriptions, but they’re struggling to turn those perceived boosts into economic bargaining power. Exceptions are maybe those few got a job at frontier AI labs.

At best, the majority of engineers are earning the same amount while paying more for AI tooling. I’m not sure what happens to them when the aggregate demand for human programmers drops due to AI automation.

That’s a great point!

> Actually we bet the other way around. History always under estimated technology.

To throw a few more nuggets, being pessimistic about aviation actually paid off. As Buffett famously pointed out in his 1999 talk, there had been zero money made from the aggregate of all stock investments in the airline industry in history.

It was a monumental technological success, but turned out to be a financial bottomless pit.

The author uses airplanes as an analogy. I want to add a few historical facts here:

- First, airplanes in the first 30 years were extremely rough and dangerous. When Dole hosted a contest for the first transpacific flight in 1927, 6 out of 8 teams failed. Ten people died as a result.

- Second, people made a lot of silly predictions about this new technology. The New York Times extrapolated that people would soon start commuting to New York using personal airplanes.

- Third, people made capital investments based on these wild guesses. For example, the architects of the Empire State Building installed a huge spire to prepare for the upcoming age of airborne traffic (intended to be used as an airship docking station!)

The bottom line is that predicting the trajectory of a new technology is often impossible. As a society, we are currently in the experimental phase of LLM-based AI, testing all the possibilities one by one. Who knows which path leads where.

I’m curious what that blight red smoke flickering around the star is. Some kind of interplanatory gas?

This discussion reminds me of “Only the Paranoid Survive” by Andy Grove (Intel co-founder). In this book, he talks about “a technological inflection point”, which is a time in the life of a business when its fundamental changes.

Intel had one in its early days. In the 1980s, HP notified Intel that new Japanese memory chips were way more reliable and cheaper than American products. Intel first denied vigorously, but then made a strategic decision to get out of the memory chip business, and enter into the new and unproven microprocessor business. This decision allowed Intel to survive the 80s, while Mostek and others basically disappeared.

What happened to 15,625 shares you did exercise? I guess they are worth over $1.5 billion as of today?

Full text of the treaty: https://corum.ktb.gov.tr/EN-61494/kadesh-war-and-peace-treat...

It reads more like a military alliance agreement between kings:

1. Reamasesa-Mai-amana (King of Egypt) and Hattusili (King of Hatti) are best friends forever.

2. If a foreign country invaded one of us, the other one must send his troops.

3. Ditto with civil wars. We help each other.

4. If a high-profile person flees from Egypt or Hatti, send the bastard back.

5. If people flee from Hatti to Egypt, order them back. Do not retaliate against returnees, though. No pulling eyes/tongues. No cutting feet/ears. No destroying wives/houses/kids. Please.

This isn’t correct. For example, when Edmond Halley created the world’s first life table in the 1690s, he estimated that approximately 5% of his city’s adult population was >70. Getting old is not new.

People also underestimate how dangerous those humanoid robots are. They are hilariously breaking every OSHA rule written by human blood.

Many contemporary humanoid robots do not even have an e-stop and this is why you see clips where people frantically try to stop an unstable robot by grabbing it. At the moment, it’s a solution searching for a problem and blood.

No, sociologists use the term involution to describe Asian rice farmers. These farmers poured an increasing amount of labor into their tiny patches of land to extract the last bit of productivity, yet despite their hard work, they remained poor.

The younger Chinese generation views their economic situation much in the same way: forced to overwork, with surprisingly little personal gain. They are typically not capital owners at all.

Note: the term is sometimes translated to “内旋” (“inward spiral”) in Japanese, which I believe conveys the nuance better.

All right. Thank you.

I understand your logic ("the high CapEx is the moat"), but on the other hand, isn't it be a bit like multiple high speed railway systems trying to connect San Francisco to Los Angeles?

And there are three internal players chasing the same goal at the moment (OpenAI, Anthropic and Google), and two others (Deepseek and Alibaba/Qwen). What will prevent them from cutting the price floor each other?

Looking from a different angle: Microsoft has been able to maintain its monopoly because it was/is a huge pain for companies to switch the operating system, but do LLMs have that stickiness?

> Obviously Photobucket completely failed to properly monetize

IIRC Photobucket actually made a good amount of money through their advertising business unit ("Give free storage and get paid by ads" was their business model). They were acquired successfully by Fox for $300M in 2007.

Ontela was a photo-uploading app provider in the pre-iPhone era. When Fox decided to spin out Photobucket (as a fallout of the MySpace debacle), the two companies got merged.

Can I ask what is your opinion about their core CapEx, i.e. model training?

The general trend I observe is that the "shelf lives" of large language models are really short. It costs $1-10 billion to train cutting-edge models at the moment, and they only really last 6 months at best.

There seems to be very little brand loyalty too. Whenever a shiny new thing comes out, people just switch over, which implies that they constantly need to fight the time decay.