Onavo

Must make building backpropation algorithms really easy.

Material Design engineers furiously prompting Gemini to copy it

Will ChatGPT dream of Bloom's 2 Sigma problem?

https://en.wikipedia.org/wiki/Bloom%27s_2_sigma_problem

This behavior is actually specific to ChatGPT because they lost a music copyright lawsuit in Germany. They would refuse to output music lyrics too but they would happily do analysis on lyrics if you supply them. I suspect there might be a guardrail model involved here.

And presumably a bunch of employees left and no longer has access to it.

But I bet the landing page was made with AI assistance.

I agree, it's the old academia play. When they can't innovate on the model work, they make a lot of noise about AI safety, gatekeeping, humanity extinction etc.

In your previous thread the solution was either WAL2

https://news.ycombinator.com/item?id=40688336

Sigstore?

Does it support pointer arithmetic?

How nano are we talking about here? A single transformer head and a few dense layers?

Alongside the original snapshots, we’ve included a few AI-powered modernization attempts that demonstrate what’s possible—getting this 1990s-era C++ and MFC code building with current Visual Studio tools, connecting to modern IRC servers, and running legibly on today’s high-resolution Windows machines.

Given that MSFT is all in on Rust and WinUI now, maybe they can try doing a full port similar to Bun using Copilot. Anthropic has been milking their Bun port attempt for as much as they can.

I disagree with the approach. It's a good approach for limited domain problems, but not for general purpose. Take something like this where you will need to be able to refer to wikis and research and ask questions on Reddit and Discord to optimize playthrough, none of the goal conditioning will be useful.

https://www.feed-the-beast.com/modpacks/125-ftb-evolution

I think a properly fine-tuned VLA with access to tool calls can scale way better.

Microsoft can probably scale Minecraft up to support infinitely large sized worlds if they want to. Minecraft lazy loads the world in chunks so you there's no real limitation (aside from machine digit sizes)

I think what OP means is that the "minimum viable product" for a daily use LLM is probably somewhere around e.g. GPT 4o's level of intelligence (YMMV). Below a certain threshold, you are better off using specialized machine learning models rather than general purpose LLMs. It's very difficult to get that level of intelligence fully local on a mobile device without streaming to the cloud.

I wish there are easy ways to figure out the maximum current and wattage supported by a cable. So many cables don't label themselves except on the box!

Is this truly more secure though? The host can still see your data.

The way I see it is that it's a structural failure of the licensing system and the FAA. Imagine after a few high profile celebrity car crashes, the Feds decided that new car designs after a certain year carries manufacturer liability and everything before has no liability. If you want a car at a reasonable price, you have to go assemble your own from pre fabbed parts. It forces the industry to become all-or-nothing with no reasonable room for nuance or debate. That's basically lazy lawmaking. The powers that be are hoping that general aviation as a category goes away completely. (And the useless AOPA is complacent, support EAA folks)

I would prefer a twin high wing with the existing TP-R90s. Unfortunately manufacturers don't like dual high wings :/

Their bandwidth costs are overpriced yes but in terms of storage you are paying for RAID3 replicated across three data centers (configurable) and high availability. The engineering behind it is also formally verified with strict theoretical bounds on data loss. $48*3 = ~$150 is within the same ballpark if you factor in the managed services and cloud overhead.

Ah, a Skywagon enjoyer?

It's worse now, people are doing "51% Build Assists" (most famously Vans and Sling aircraft) to save money where the factory would give you an IKEA plane kit and you will maybe put in a few screws and decals and install the avionics panel and you will sit in the room and watch them assemble the rest of the plane "under your supervision". Then the whole thing will get penciled whipped as an Experimental Owner-Built aircraft and if anything goes wrong it's entirely the fault of the owner and not the factory, because the layman hundred hour PPL pilot is supposed to be a master of aerospace engineering.

I think MOSAIC is the only thing the FAA got right in the recent years when it comes to general aviation.

A better way would be to switch to mogas and authorize cheap fuel system retrofits and maybe mandate dual FADECs on engines similar to the ADS B requirements. Your modern general aviation pistons can easily handle mogas.

An even better way would been to encourage the migration to turbines and deprecate pistons entirely. Jet A (diesel) for the win.

The new turbotechs regenerative turbines are a lot more reliable (3000 hours TBO).

https://cubcrafters.com/c/2026/07/press-release-cubcrafters-...

Pistons suffer from too many reliability issues (especially if you don't have a MX team on call like a commerical operation).

I am not sure why the aviation industry (and the FAA in particular) is obsessed with keeping around almost-a-century old lawnmower engines (Yes your Cessna Skyhawks, if you trace them back to their predecessor, are from the WWII days). Learning to "lean an engine" never made me a better pilot and there's a reason FADECs are used everywhere else.

It's a very big problem because most of the big general aviation players (Cirrus, Diamond, Continental, a bunch of others) are snapped up by Chinese conglomerates (they don't have general aviation in Asia but they have a growing number of flight schools, not sure how that works) now and they are hardly known for their expertise in piston engines. People buy Chinese EVs, nobody buys Chinese ICE cars. Diamond owners had to ground their planes for over 2 years because Diamond couldn't figure out how to do a Mercedes OEM license-built piston head properly. The Chinese-built pistons kept cracking and required boroscope engine inspections every 50 hours (notoriously expensive and not covered by warranty, they only managed to solve the cracking problem recently). So much for supposed Chinese manufacturing prowness.

Imagine buying a fancy Porsche car and only to be told that you have to keep it in the garage most of the year because the car company got bought out and the new owners don't know how to produce a proper piston. Owners would riot. Sadly general aviation operators are trained to take it up the ass like a good little pilot because the dinosaur FAA (for whom general aviation is their lowest priority, about the same importance as hot air balloons) mandated that the OEM's word is literal law and if it's grounded then too bad.

Is Python about to have its Project Loom moment?

Why sell the farm? Somebody needs to keep up the fight against Big Agriculture.

To let the model know when to call them, you send a list of it to the model as part of the context. Each MCP contains a description and sometimes each tool contains a description.

Claude code is sandboxed by default no?

The hard part for B2B is the sales not tech. You are better off wining and dining a sales guy as your co founder than doing more vibe coding.

Look online for Kx systems history of IP enforcement and legal threats. They are well known in the quant community for essentially being the Disney of time series databases. Nice tech yes, but the people behind it hold values that are typically diametrically opposite of hacker ethos. Good riddance.

I hope OP can survive the lawsuits though, the K authors are infamously litigious.