I like Chapter 1 of Evan Chen's An Infinitely Large Napkin for some more theory.

https://web.evanchen.cc/napkin.html

I fixed a malfunctioning refrigerator by replacing the control board, which happened to be a PIC16-based device used by many brands. This design is as close to "universal" as you can get, and this generic board was around 20% the price of the official replacement part.

https://www.ebay.com/itm/152763501102

That's because many different brands are actually made by Whirlpool. They're like the GM of the appliance industry.

How were the contributions by Richard Caley handled? "The legal reality was harsh: Richard’s contributions to Conquer couldn’t be relicensed. The university couldn’t help contact heirs due to privacy laws."

Author of the article here. Richard's contributions remain in the codebase but under original terms. We documented his legacy as a person, and that is explained in the README of the repository.

The notion that everything had to be relicensed under the GPL “so it could be properly preserved and packaged for modern Linux distributions” seems pretty silly.

Only if you haven‘t dealt with licenses. This software was published without a license.

It's a wonder the software's authors, users, and distributors have evaded dire legal consequences all these decades.

It's a ticking time bomb that distributors are steering clear of for good reason. You are correct that current copyright law is pretty silly, but that silly law is frequently enforced by serious courts.

The R3000 had a moment in the sun, it was used in everything from SGI and DEC workstations down to the Sony PlayStation.

Should've made it into a long-playing laptop, imho...

How does this sort of thing work from a technical perspective? Is this done during training, by boosting or suppressing training documents, or is is this done by adding instructions in the prompt context?

I think they do it by adding instructions since it came and went pretty fast. Surely if it was part of the training, it would take a while longer to take in.

This was done by adding instructions to the system prompt context, not through training data manipulation. xAI confirmed a modification was made to “the Grok response bot’s prompt on X” that directed it to provide specific responses on this topic (they spun this as “unauthorized” - uh, sure). Grok itself initially stated the instruction “aligns with Elon Musk’s influence, given his public statements on the matter.” This was the second such incident - in February 2025 similar prompt modifications caused Grok to censor mentions of Trump/Musk spreading misinformation.

[1] https://techcrunch.com/2025/05/15/xai-blames-groks-obsession...

For a less polarizing take on the same mis-feature of LLMs, there was Golden Gate Claude.

https://www.anthropic.com/news/golden-gate-claude

The creator of this game is a frequent contributor on this site.

https://news.ycombinator.com/user?id=WalterBright

Is that the same Empire? The version I've seen has no BTUs. Just basics: random continents, cities produce military units, units require different numbers of turns for production, etc.

However, I saw a modern-ish decendent called Empire Deluxe, so maybe the original had more features?

The version I played all those years ago definitely had BTUs. I remember planning ahead of time what the most efficient use of them would be, before actually doing things between updates; not to mention setting alarms to wake me up, log on, and use up the (otherwise wasted because I was at max) ones at times in the night, if I was doing something, er… sneaky :)

TIL :)

LMGTFY:

https://translate.google.com/?sl=en&tl=fr&text=That%20jumbo%...

I woldn't go so far as to say it was "prophetic". Contemporary DEC PDP-8 (OMNIBUS) and PDP-11 (UNIBUS / QBUS) systems have a similar approach to "interoperability", where cards for peripherals were also mapped into the machine's address space. It was great that Woz saw the utility of this and brought it into the homebrew/microcomputer design.

I understand that Steve knew about that but trying to create an inexpensive computer, working with hard trade-offs makes his decision very wise and smart. We can know about complex architectures but it might be very difficult to copy them in cheaper devices.

I think there's an error in the Position Independence section:

    start:
        auipc a0, 3
        addi a0, a0, 4

The text says that this should result in 0x3004; was this example intended to be

    start:
        lui a0, 3
        addi a0, a0, 4

No, because `lui` results in an absolute address, not a position independent one.

The `auipc/addi` sequence results in 0x3004 + whatever the address of the `auipc` instruction itself is. If the `auipc` is at address 0 then the result will be the same.

Exactly, but the text has the same instruction sequence twice and the parent correctly indicated that the first copy should have used "lui" to illustrate the problem you mentioned and the second copy does use "auipc" to illustrate the fix you mentioned.

Oh, fair enough.

This copypasta has been fixed.

"Linux on the Desktop" is great. I've been using it since 1994. "Linux on the Laptop" sucks- I just want my laptop to sleep and awake properly, without draining the battery. I'm old enough that I'm done spending time twiddling kernel parameters in an attempt to get all of the onboard devices working, including sleep.

> I just want my laptop to sleep and awake properly, without draining the battery.

To be fair this is _also_ a massive problem on Windows too, because of Windows Modern Standby encouraging laptop makers to replace ol' reliable S3 sleep with the terribly broken modern standby stuff. Macbooks and certain Framework models are the only laptops left with reliable sleep.

Old video but nothing's really improved since: https://www.youtube.com/watch?v=OHKKcd3sx2c

Not a laptop, but steam deck sleeps perfectly. So it's not a Linux problem, but a laptop problem.

Meanwhile, Microsofts cannot even get their own Surface line with Windows on them to not wake up while in your backpack.

Hell, my Pinephone Pro slept perfectly well as well. Beta-level hardware without a major company behind it, and sleep still worked reliably.

At this point I expect sleep to work better on Linux than Windows machines.

I just want a laptop that has almost zero latency between the CPU and RAM and at least 300GB/s RAM bandwidth for data science. Not much choice there, unfortunately.

“Linux on the Desktop” is a 30 year old meme about this being the year that people will leave Microsoft en masse and install Linux.