Centigonal

5,456 karmajoined 14 anni fa

{my_username} at gmail

comments

Yes, I am. I think Chatto looks great, and I use AI in my job every single day. I've been following OpenAI since MuseNet. I'm uncomfortable about the coming Kuznets curve for AI, the wealth concentrating effects, the energy use, the stunting effects on human intelligence and connection, and the abuse of training data associated with AI. I am that "let's take a nuanced approach and find out how we can get most of the good without most of the bad" person. I think people like me are probably the majority stance when it comes to AI use among tech people. I also think that we'll be steamrolled by moneyed interests and society-level incentives in the short term, and I don't really see a way around that.

In the absence of a way to win with nuance, I think it's just nice to have people walking around who live rigidly by their principles as an example of what is possible.

l’altro ieridiscuss

If they fixed it, then it wouldn't be SWE-Bench Pro anymore, right? It'd be "SWE-Bench-Pro-Fixed-OpenAI." I think it's better optics for the independence of the benchmark if the OpenAI team lets some third party do the fixing and release the improved benchmark.

...Although OpenAI did exactly that when they released SWE-Bench Verified, so maybe I'm talking out of my butt here.

l’altro ieridiscuss

Good luck! The world needs more rms types (specifically referring to people who adhere strongly to moral principles over expedience)

3 giorni fadiscuss

That was intentional - I think Office (or, uh, Microsoft 365 Copilot as it is now called) is frustrating mass-market mediocre software! That said, it consistently lets me do my job, while LibreOffice is often unusable for my purposes.

4 giorni fadiscuss

For 2 and 3, office software and OSs have strong network effects and up-stack effects, just like CPU instruction sets.

Also, I'm sorry, but OSS office suites compete with Office and GSuite the way grocery store frozen pizza competes with Domino's and Papa John's. Quality and completeness of execution matter a lot in that category.

5 giorni fadiscuss

It's in the press release.

> Compulsion Games and Double Fine Productions will return to management and transition to independent studios with their IP, catalog, and runway for their next games. Ninja Theory and Undead Labs have entered terms to join new ownership with funding to complete and grow Senua and State of Decay 3. In France, Arkane’s management is beginning required consultation with its Works Council to review potential strategic options.

10 giorni fadiscuss

Looking forward to reading something very similar to this in 10 years about the Discord servers and Insta DMs of the 2020s.

11 giorni fadiscuss

There are some nice curb-cut effects of dark-sky lighting:

- The circadian rhythms of birds and nocturnal creatures are less severely affected

- Amber lights produce less insomnia

- The dumb streetlight across the street doesn't blind me from my bedroom window

11 giorni fadiscuss

Larry Sanger has been grinding an axe against Wikipedia for nearly two decades. Some of his arguments are convincing, but it's hard to take them seriously when they're coming from a man who has tried repeatedly to undermine the site since the mid 2000s.

12 giorni fadiscuss

One of the arguments Hank makes in the video is that SpaceX is (via starlink) rapidly occupying large portions of useful LEO shells, which crowds out future competitors or users of that orbit (i.e. you can't put more satellites into the orbit without risking collisions, especially satellites that aren't part of the existing constellation), and that the natural consequence of not regulating orbital space in some way would be to lock in the first movers in an orbital shell as the only organizations that have access to that orbit.

12 giorni fadiscuss

Hank Green did a video recently advocating for an "orbit value tax" -- like a Georgist Land Value Tax, but for orbits. This tax would, among other things, help fund orbital cleanup and internalize the externality of polluting orbital shells. It's an idea that deserves more discourse IMO.

Here is the video: https://www.youtube.com/watch?v=VLjW6zuYmos

12 giorni fadiscuss

"Rocket Lab acquires Iridium" sounds like a notification out of Sid Meier's Alpha Centauri or Anno 2205.

12 giorni fadiscuss

That iMessage rendering looks correct, just typeset differently.

15 giorni fadiscuss

There is an active debate on exactly this topic when it comes to whether or not to excavate the tomb of Qin Shi Huang.

https://en.wikipedia.org/wiki/Mausoleum_of_Qin_Shi_Huang

26 giorni fadiscuss

jqwik users want the sort of things that jqwik enables, too.

26 giorni fadiscuss

You're right, I don't want the software on my devices doing things the user doesn't acknowledge or want, that has potentially destructive/negative consequences. Down with that sort of thing!

So when are we nixing Widevine, EasyAC, carrier locks on phones, and TEEs that the user can't look into?

26 giorni fadiscuss

I don't know if I agree with this, especially in the context of animation: https://www.bloopanimation.com/the-art-of-smear-frames/

29 giorni fadiscuss

The model is trained to treat system messages differently from regular user/assistant messages. Most models are trained to only expect system messages at the start of the conversation. This is changing now.

29 giorni fadiscuss

For Deepseek V4, The main effect of raising the reasoning effort is to add a little section[1] to the end of the system prompt that says "BTW make sure to think really hard! :)"

If Anthropic's models work the same way, then changing reasoning effort would break the cache because the API has to modify the system prompt given at the very start of the context and rerun the whole thing through the inference server.

This kind of limitation is one reason Opus 4.8's mid-conversation system messages[2] are actually a pretty big deal (if they actually work).

[1] https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash/blob/ma...

[2] https://platform.claude.com/docs/en/build-with-claude/mid-co...

30 giorni fadiscuss

This is true. Amazon Seller Central, AWS Partner Central, and Glue Databrew are all Amazon productivity-style apps, and I hated every minute of using them. Quicksight is okay these days, but it took them a decade to get there. They really struggle outside of devtools and consumer-facing web experiences.