croemer

Parent fixed the error I pointed out since making my post by dividing by a thousand instead of changing the units so now my comment looks wrong.

It used to say 2800 GB/s which is obviously bonkers.

This has strong signs of LLM writing, potentially Claude.

You probably meant MB/sec not GB.

No, they don't, they can also be defaulted on/written off. Open weights are there to stay, plus research, plus the GPUs, just the power needs to be paid.

There's a limit to price hikes due to competition, and open weights can't collude so closed weight companies don't have that much wiggle room to raise prices unless they are much better than open weights.

I was skeptical because this is AI written but Claude Code with Sonnet 5 managed to reproduce it convincingly. Sure I didn't manually verify but it's a lot more trustworthy to have your own agent verify than just trusting a blog.

Can you run Claude Opus through the same Pydantic harness and add the cost to the benchmark result table? An isolated price is meaningless.

The tool is handy, managed to confirm that dash and ash behave differently.

But why should I trust your tool if your writing is wrong?

They are not the same. For example, `[[` apparently works on Alpine's ash but not on Debian's dash. BusyBox ash has some bash compatibility options enabled in Alpine.

It's ironic that a post about shell differences glosses over this. Maybe not surprising that slop is sloppy.

Inconsistent: claims /bin/sh on Alpine is dash in one paragraph, BusyBox ash in another.

Smells a lot of AI writing.

It does:

The remainder of this post is an experiment in agent engine optimization, a genie-generated description of YAGNI

Genie-generated means AI generated

The dollar amount is meaningless without comparison - and no other model has a price tag. Sloppy article.

They should also at least run Opus through the same Pydantic harness they used for GLM. As is, it's apples vs pears.

Where's the cost per vulnerability for all the other models than GLM?

Also, without code this isn't very trustworthy. Could all be made up as well.

Have you tried switching off custom stuff that might get added in beyond the prompt?

Fable (through claude.ai) refused all my prompts even "How many Rs in Strawberry" claiming it was related to biology or cybersecurity.

I had to switch off memory and my custom instructions to get it to stop refusing. It turns out if you even mention that you work with bioinformatics software you get blanket refusal.

I use chore quite a bit in my human written commit messages.

Tandem x2 alarm fatigue is insane. Constant vibrations about every useless thing.

Mylife Ypso with CamAPS FX though is really good. Just does the job. Alarms are centralized in the CamAPS app and you can silence them - something that x2 never let you do.

Fellow T1D here. The same pump tslim X2 also failed on me 3.5y in, also on holiday (I was in Morocco). I had backup pens with me so no big issue.

I've now switched pump to my Ypso mylife with CamAPS FX algorithm. It's so much better than the tandem! Much fewer useless alarms, the algorithm is way better. Downsides are manageable: reservoir is only 160 units (instead of 300) and one needs phone to be always on (good motivation to keep it charged).

TLDR: Products affected: "All dedicated servers and cloud plans at all locations are affected by this adjustment. However, web hosting products, managed servers, servers from the Server Auction, IPs, storage products, Load Balancers, Volumes, Snapshots, and Object Storage are not affected."

No info on how much prices will go up: "We will provide further details on the new prices on our website during the rollout."

More than 15 min after it started an incident is finally posted: https://www.githubstatus.com/incidents/xy1tt3hs572m

Your website bio and LinkedIn don't match at all. Is the LinkedIn link on your website wrong? Update: yes it is. This is the correct one: https://www.linkedin.com/in/logan-nye

You're doing too much vibe coding and not enough checking/testing.

LinkedIn link on your website points to: https://linkedin.com/in/logannye

Website bio: https://www.logannye.io/about

We rewrote Nextclade in Rust and are very happy. Works nicely both for CLI and client side browser with wasm.

https://github.com/nextstrain/nextclade

Those are all the tests for alignment. They don't even check the alignment is correct. Just that there are no errors. This is a joke: https://github.com/logannye/rosalind/blob/main/tests/alignme...

Looks like total slop to me. All code in one commit, then a bunch of commits polishing the Readme.

No release, no updates in half a year.

If it is open-source, you will see a meaningful activity - stars, forks, contributors.

That's not true, I'm quite sure most repos on GitHub have neither many stars, nor forks, nor multiple contributors.

Issuance was stopped almost 2 hours ago: May 8, 2026 18:37 UTC.

Reads at least partially like LLM writing, for example:

When code production gets cheap, the cost doesn't disappear. It migrates.

It was true then. It is unavoidably true now.

Staccato sentences, em dashes, word choice, the headings, "X is not Y. It's Z.":

The legacy of the book has almost nothing to do with the satire. It has to do with the language.

Or this also reads very LLM like to me:

None of the cruelty is gratuitous, exactly. It is dramatised exhaustion with the genre.

You can suggest changes on the talk page.

Wikipedia provides a structured process for subjects to suggest changes without directly editing articles. The mechanism centers on the article's "talk page", a discussion area attached to every Wikipedia entry where editors coordinate improvements.

https://en.wikipedia.org/wiki/Wikipedia:Editing_a_page_about...

Someone should try to replace the neurons with urand and see if the chip can still play Doom, in the spirit of the qday prize winner.

At this point, it looks just like an experiment. It's not a definitive "were going to switch".

I think people here are reading too much into it.

Presumably stock market valuation is grounding?

Also, eBay shareholders can vote down the acquisition if they don't think the deal is good for them.