abdullahkhalids

A harness [1] was developed by Terrence Tao and some collaborators to prove mathematical results. It has since then been used by others with positive effect. Can someone critique the structure of this harness? I don't know anything about this stuff.

[1] https://github.com/1stproof/batch-2/tree/main/batch-2-submis...

> is there a real benefit for a Sandstorm grain over a docker style Linux container?

Linux containers are meant to be used by those with significant software engineering skills. Sandstorm was designed to be used, once installed by someone else, by grandma.

Both Germany and Pakistan installed roughly 17 GW of new solar in 2025 [1,2]. Pakistan's GDP per capita is about €1500.

[1] https://janrosenow.substack.com/p/pakistan-the-solar-revolut...

[2] https://eu.globalsolar.store/news/78/germany-installs-175-gw...

If you increase the date range to 12 months, you see that (with "likely human"), linux usage jumped from 3-4% to 5%-6% in the last 1.5 months. Something changed.

ChromeOS jumped down several points in the same time period. So maybe some update of ChromeOS got misconfigured?

> how then to previous other posts that point at them, or reply to them, get updated?

You have outlined two scenarios. One where the old posts are deleted, and in the old posts are moved to the new host.

In the former case, the "referencing posts" start pointing to 404. Maybe in the latter case, we can also just have the referencing posts also point to 404.

A partial solution is better than a full solution.

What languages does it support?

Restriction is a widely used technique in both art, and life. Consider black and white paintings, or generally paintings which only use shades of a single color? Why are they not using all the colors when they exist?

What about people who do challenging athletic endeavors with limitations? The ice guy climbed up to the base camp of Everest in just his shorts and slippers. Why not wear more?

What about minimalism as a way of life? Why not buy more if you have money?

This is just how humans are.

Large publishers are optimizing their overall revenue/profit, not the revenue/profit per book. Given that large publishers dictate a significant fraction of what appears on bookstore shelves, how concentrated publicity drives of new books/authors drives their sales at the expense of old books/authors, and how the emergence of second hand books outside their control drives sales in different categories, its extremely likely that the advantageous strategy often is to not print older titles even when there is demand for them.

There is an easy way to mostly replace email over a decade, with the NewEmail protocol

NewEmailServers are able to send and receive messages using either the Email or the NewEmail protocol.

Before sending a message, the NewEmailServer checks a directory (similar to DNS) to see if any of the recipients are on another NewEmailServer. If so, the message is sent to those recipients via NewEmail. Others get it by Email.

Completely backward compatible. Once google/microsoft shift to NewEmail, the rest will follow quickly except some legacy servers. All that is needed is a standards body.

There are three clear benefits.

1. Answering where humans come from is probably one of the the oldest question we have had. All religions try to answer this in one way or the other. I say, trying to come up with scientific answers is extremely valuable.

2. We feel awe at the wonders of nature, not much different from the awe that we feel when we see great human art. Tax dollars as well as charity are used to subsidize art often. This is similar, though there is more money involved here.

3. The problems one sees in different domains are always different. Funding agencies give money to the "directly economically useless domains" because the workforce that is trained and technology that is developed can then be applied to the "directly economically useful domains". The support money to the former is much smaller than the support money to the latter, but just enough to inject fresh ideas and creativity to the latter.

It's important to note that even if these open models are distilled, they are showing genuine improvements in their architecture, which enables inference costs to be several factors below what equivalent closed models have.

The interesting question is: will Anthropic release a Fable like model with an architecture similar to Kimi, and get the inference cost gains? They should surely beat Kimi because they can internally distill as much as they want.

Doors aren't humans either, yet we design their handles and locks to be graspable and manipulable by humans.

The purpose of technology is to serve humans. Therefore, technology must conform as much as possible to human sensibilities rather than vice versa.

I will accept a 5% drop in benchmarks for a model that talks to me like a human.

Mistakes in proofs are relatively common, but such a mistake doesn't automatically mean that the proof is entirely wrong. Many times, the mistake is just in the exposition, and can be fixed easily. Other times, the mistake is fixable and the fix is apparent. Perhaps the author forget to treat a relatively trivial edge case. In the first two cases, the student would likely just pass with minor corrections to be submitted soon.

Sometimes, of course, the proof is just wrong. That is the dangerous case, which will cause either major corrections or failure.

Funnily enough, you will have to pretty accurately copy the laws of our universe to make an Apple pie. Even small changes will change the properties of atoms, and hence all of chemistry and biology. And there won't be any apples or wheat.

"How does Kimi Work protect my privacy" can be interpreted in two ways. The first is "How does Kimi Work protect my privacy from Moonshot?" They did not answer this question. The second is "How does Kimi Work protect my privacy from other third parties?" They are answering this question by saying, we won't upload your files to some third party server without your consent.

How many free hosts from 20 years are still alive? Not many.

Dream would be a combination of AI models that are smart like a human transcriber. One can just tell the model what mistakes it has made (e.g. "No, you wrote XYZ but I meant WXY"), and it is intelligent enough to realize when it is being instructed and when it needs to transcribe exactly what I say.

This is pretty cool. The dream would be that all editors (including brower tabs) have an api interface which allows arbitrary LLMs to modify the editor content. So you can indeed look at different windows as you talk, but the editor keeps getting updated live, so you can go back and see where you already at.

In fact, a few apps I have tried do exactly this inside the app themselves. There is a live textual field that displays whatever the model thinks, and this display constantly goes back and fixes earlier words.

So I know it's possible. I just want to integrate this with the paste-at-cursor feature that these apps have. I imagine the app would have to create a virtual keyboard and use backspace or arrow keys to go back and change things.

For anyone looking to build on top of this. I have tried a few different STT systems, and they accurately capture what I am saying. Unfortunately, they don't support the reasonable workflow

I want to open an office document, for example, and start talking. And I want the software to continuously type what I am saying at the cursor with minimal latency. The continuous part is crucial. Many software will paste whatever I said after I have stopped recording, but that is not useful.

I would bet the legged robots are advertisements, and when a factory comes knocking they are provided the option of buying cheaper robot arms that are stationary or on wheels.

Robotics and textual AI have very different risk profiles. Textual AI can just be provided to users even if they are completely crap because some users will get some usage out of them.

Robots can only be released once they are good enough to actually do anything, and have passed safety certification. So, not seeing robotic products out there is completely natural. But they are not far behind LLMs, and they will very suddenly take over the the factory world as soon as they cross these performance and safety thresholds.

Unfortunately for workers, these type of actions can only term delay their replacement only by a few years, if that. The only long term solution here is to go bigger and reconceptualize the entire economic system, a system that takes into account the new realities that AI has created, and provides justice to all.

> Lime is more breathable, workable and sustainable. The only thing is, it requires maintainance

What sort of maintenance does it need?

The confounding variable I think is most important is that many people have an internal clock that is shifted or lengthened relative to other people in the same house.

If others in the house prefer to sleep from 10-6 and you prefer to sleep from 12-6, but others start making noise a 6, your sleep quality in the last two hours is destroyed. Then over time, it just results in poor sleep regularity, as you cycle between exhaustion and trying to sleep according to your internal clock.

An interesting solution would be for these AI companies to train a few different versions of these models, all with different speech characteristics. Then, when you start a conversation, you get a random version.

Just today, I got frustrated with the language. I searched around, and in my Claude Instructions I put in Ref [1] (translated to English). It is certainly better phrasing (though still quite annoying), but I don't know if this makes the output technically worse in some way.

[1] https://github.com/hexiecs/talk-normal/blob/main/prompt-chat...

I wouldn't expect this to be better tasting than a regular soup/stew. If you keep eating it, new stuff added today has been almost completely consumed in a week.

I would eat it out of respect for the craft and the values that are being preserved.

In text, you have the advantage of putting things in any order, with the assurance that the reader knows exactly how long the message is.

Always start your email to a busy/important person with exactly what you want from them. Within the first two sentences. Then peel the onion, giving details from more important to less important.

This gives the reader opportunity to stop reading at any point having decided to help you or not, with the least waste of their time. In many case, the person can skill your description of your work, and just tell you what to do.