blackbear_

As we all know, oeer review is not a very reliable signal about the veridicity of the claims, it is merely an approximate quality gate indicating that a study was done following appropriate standards. There needs to be more awareness of this. Nonetheless, in some fields peer review have been public for a long time, and more and more journals are making peer reviews publicly visible. This is a good trend that we should all encourage.

I also think that it sounds like a good idea to hold scientists accountable for the quality of their research, but moving away from fraud (which should unquestionably be punished) towards merely "low quality" is a very slippery slope. Who determines when the quality is good enough and under which criteria? Citations and follow-up research already are an approximate signal, why is that not enough? Proving a study false does not automatically make it low quality. How much money and resources would these additional measures cost? Would the savings from fewer "bad quality studies" offset these costs? How many fewer "good quality studies" would be prevented by redirecting these resources.

Totally agree with this framing: trust in public institutions (scientific, governmental, etc.) is a common good that needs to be promoted and safeguarded.

Identifying and punishing fraud is a part of it, whistleblower protection another, modernizing the system so that graduate students are not so dependent on individual advisors yet another idea. More and more personal contact between the public and scientists could be useful too, for example via science fairs, meet & greets and so on.

Yup, that is also how it works with scientific research: Once an article is published, anybody is free to read it, criticize it, reproduce it. That's because the government stays out of it: private individuals can use things like Google Scholar and Google Search to find other people's reviews and follow-up studies to better understand the quality and nuances of the original research. Consensus is an important part of science just as it is for reviews on Google Maps. Is consensus sometimes wrong? Yes, no doubt. Is it useless? Nope.

> If people can't define it, then that seems to me like a valid criticism of the scientific method.

Please read the hamburger example once more.

This is unnecessarily cynical.

> The core problem is people conflating science with Science™.

You fell into the same trap by criticizing scientific institutions and the human spirit in an attempt to criticize the scientific method.

> What exactly is the scientific method?

What a ridiculous argument. What exactly is a hamburger? Just ask around and you will find that nobody agrees if a veggie burger should be considered a hamburger, or if you can replace the bun with lettuce and still call it a hamburger. And even if you get people to agree what a hamburger is, nobody is checking that the recipe is followed to the letter. Therefore, real hamburgers do not exist, and nobody can claim in good faith that they are making a hamburger. In fact, why don't we sue McDonalds for their fraud hamburgers?

> Science isn't self correcting.

The very fact that fraud is identified is clear evidence that is is not true. You are going to need to sharpen your rethorical devices if you want to argue against hundreds of years of scientific progress, including for example everything that enabled us to argue about this in front of our devices while barely 150 years ago people had to use candles to see each other after dinner. Mind you, not all of this was developed by industrial research labs, in fact the origins of very many modern technologies lie in academic research labs pursuing ideas that are too risky for a commercial entity to bet on. And of course there are plenty of examples of industrial fraud happening despite "managerial intervention" (lol).

Reality is more complex. Not many things, if at all, exist in this world under their platonic ideal form. That does not make them any less real. Yes, we all know that humans are not perfect, sometimes are selfish and chase money and power, not only in academia but also in business, politics, and so on. Wherever humans are involved this sometimes happens. But saying that nothing good ever happens because of this goes against everything one can see when looking out of the window. Scientific progress is real, and needs both industrial and academic labs.

Everything is a linear model... Including ANOVA. I find it much easier to condense everything under this framing instead of remembering tens of tests and ad hoc rules. Frequentist statistics always felt to me like an uncohesive bag of tricks anyways.

https://danielroelfs.com/posts/everything-is-a-linear-model/

This was abundantly clear when people, even on HN, were upset about the bureau of labor statistics revising their numbers tendentially downwards, probably confusing the notion of statistical bias for that of political bias.

If the figures are biased, just estimate the bias and correct for that, what is the big deal they said, as if the bias variance tradeoff was not a thing.

Labels did work in reducing harmful behaviors:

> Participants drew on a range of risk discourses in framing their abstention or moderate drinking along the lines of health, wellness, wisdom and productivity. They reiterated social constructions of heavy or regular alcohol use as irresponsible, threatening and potentially addictive.

https://onlinelibrary.wiley.com/doi/10.1111/dar.13709

Having similar thoughts at times. What are you reading nowadays?

The GPT3 paper is a good starting point

Language Models are Few-Shot Learners https://arxiv.org/abs/2005.14165

I also enjoyed the papers for DeepSeek and GLM for an overview of all the tricks you need to make these things work

DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models https://arxiv.org/abs/2512.02556

GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models https://arxiv.org/abs/2508.06471

There are other players in the game: the business and the market.

Good code makes it easier for the business to move fast and stay ahead of the competition while reducing expenses for doing so.

Is it because people genuinely don't care, or because the barrier to become a power user is becoming taller and taller every passing year?

> what kind of neural-net architecture and training would allow a model to handle numbers lengths it hasn't been trained on

A recurrent neural network implementing binary addition with carry could do this, and one can derive the correct weights with pen and paper without too much effort.

Whether gradient descent will find them too is another matter entirely

> I measure, test, and validate outputs exhaustively.

How do you do this? Do you follow traditional testing practices or do you have novel strategies like agents with separate responsibilities?

There is no escaping the fact that feeding addictions is a great business model.

Luckily our friends overseas have shown us the way of dealing with uncompetitive local industries: tariffs.

Whether the code is AI generated or not is not important, what matters is that it really works.

Sharing code enables others to validate the method on a different dataset.

Even before LLMs came around there were lots of methods that looked good on paper but turned out not to work outside of accepted benchmarks

Isn't it insane that any web page can run a port scan in the first place? Who wants that?

Meanwhile, running opencode in a podman container seems to stop this particular, err, feature.

I had similar thoughts, and my conclusion is that competition is an inherently unstable state of affairs: at some point somebody wins, and they will try very hard to prevent any further competitors from arising.

Indeed, competition is undesirable for all participants involved: everybody wants to win and exploit the rest for their own gain. Note that this is the only way to make competition work and result in its temporary benefits (if nobody wants to win, nobody will compete).

So there must be a system to keep the competition going and preventing the rise of a definitive and exploitative winner, and the existence of this system has to be accepted by the competitors. But why would serious competitors accept a system that prevents them from winning?

But we do ban tools sometimes: you can't bring a knife to a concert, for good reason.

It is both common and uncontroversial to put restrictions on using certain tools in certain situations for safety reasons, especially in public and crowded places: you can't bring a hammer to a concert.

As the provider of a public place, X ought to take certain measures to ensure public safety on its premises. Of course, deciding what is and in not tolerable is the crux of the issue, and is far from trivial.

What other countries could teach the U.S. about bringing down drug prices

A look at how European governments negotiate with pharma companies helps explain why Americans pay more for prescription drugs.

https://news.stanford.edu/stories/2024/11/what-other-countri...

Because US voters prefer the free market as opposed to government regulation and nationalized healthcare.

No, this would crash with numpy arrays, pandas series and such, with a ValueError: The truth value of an array with more than one element is ambiguous.

It is done, in many countries there are legal requirements to report adverse events whenever they are observed upon use

https://en.wikipedia.org/wiki/Pharmacovigilance#Adverse_even...

Generalized Additive Models: An Introduction with R by Simon N. Wood

This book is very strong on the fundamentals, while the R code is minimal and easy to follow.

And so fact checking is back in vogue. Seems pretty biased.

Two thoughts:

> how important is computing the exact gradient using calculus

Normally the gradient is computed with a small "minibatch" of examples, meaning that on average over many steps the true gradient is followed, but each step individually never moves exacty along the true gradient. This noisy walk is actually quite beneficial for the final performance of the network https://arxiv.org/abs/2006.15081 , https://arxiv.org/abs/1609.04836 so much so that people started wondering what is the best way to "corrupt" this approximate gradient even more to improve performance https://arxiv.org/abs/2202.02831 (and many other works relating to SGD noise)

> vs just knowing the general direction to step

I can't find relevant papers now, but I seem to recall that the Hessian eigenvalues of the loss function decay rather quickly, which means that taking a step in most directions will not change the loss very much. That is to say, you have to know which direction to go quite precisely for an SGD-like method to work. People have been trying to visualize the loss and trajectory taken during optimization https://arxiv.org/pdf/1712.09913 , https://losslandscape.com/

Tailoring the message to the audience is really a fundamental principle of good communication.

Scientists and academics demand an entirely different level of rigor compared to customers of LLM providers.