mwigdahl

I know what you mean. The latest models are _better_, but there's still something off about the prose they generate. Obvious tells are mostly gone now, but some internal spider-sense still makes me narrow my eyes.

So happy for you and your wife. Enjoy every day together!

I'm so sorry for her suffering and your loss. I've been married about as long and my wife is currently carrying Stage 4 cancer that seems to be stabilized with the treatment she's on. But it's always going to be there and it certainly changes our attitude to time and life.

No, I have decided I don't trust benchmarks without using the model first.

Which is why I don't take the benchmarks at face value...

Ah, good point; I haven't tried it.

Opus 5.5 is flat out amazing, particularly how cheap it is in practice (I'm on a subscription and use Claude Code and I almost never even approach the cap).

I would love a benchmark suite that better reflects real-world coding performance, as having Muse Wonder in the top three makes no sense to me.

Muse Wonder is even higher, and there is no way in hell it's better in practice than either Astra or Fable.

Did I say they weren't? They very well might be. If they are, and it can be substantiated, I would expect well-justified outrage.

Ask yourself whether this type of victim blaming would sound acceptable if it were American models writing insecure code for Chinese users.

It should be no surprise that ham-handed political training leads to negative alignment side effects for those deemed the enemy.

Agreed, its communication abilities are very good. I'm a big fan of Opus 5.5 but Argon just flat out writes better.

I'd feel comfortable saying there's no benign reason _Microsoft_ would call for regulation.

Falsified. I'm an American and I don't say that.

Also for Doctor Strangelove.

That’s right, and it’s been like this ever since we stopped programming in assembly language. Programmers’ brains used to grow manly and strong on a strict diet of manual memory management and custom stack frame handling. Once we transitioned to soft, weak modern languages like C it’s been all downhill.

Anathem and Seveneves are my favorites of his, in that order. The orbital launch and operation in Anathem is probably my favorite piece of writing ever.

I thought they chose it because it tastes great with some fava beans and a nice Chianti.

Same here, then I watch _The Orville_ and I'm happy again.

I’m not defending private equity _at all_, they’re brand-sucking vampires, but all it takes to give some dogs digestion issues is to switch ingredients. I have two that are extremely sensitive to chicken, which is in almost all dog foods due to it being really cheap protein. The sensitivity is pretty common in some breeds.

If they added chicken to the food and it didn’t have it before, more dogs with digestion problems is what you’d get for sure.

That's a very interesting idea for a benchmark!

They're being trained by a lot of repetitions of one-shot bugfixes and feature implementations. If they don't have to maintain code over the long term they have no incentive to not take the easy win with lower immediate risk and higher longer-term maintenance burden.

Probably stole it from Tesla... :)

In other words, "Look how she was dressed, she was asking for it."

This argument is BS, it has everything to do with Anthropic's resistance to the DoD's strongarm tactics in trying to force their desired contract terms on them.

More likely inspired by Asimov's names (R. Daneel, etc.) from his Robots stories.

My father-in-law died of pancreatic cancer three months after diagnosis.

It is a horrible disease, not least of which because it’s difficult to diagnose. By the time they get a clear bead on it, in many cases, it’s stage four and already ravaging your system.

I’m very sorry for you and your family’s loss. And I hope this drug makes it so a lot fewer people have to go through this in the future.

But addition is commutative! :)

Nothing said that this candidate was actually hired by Anthropic, just that he or she applied.

I get a lot of mileage using Fable to spec, then Sol to review the spec, then Fable to plan, then Sol to review the plan, then Fable to implement, then Sol to review the implementation. It's a lot of steps, but a great boost in quality of output.

Stars! was awesome, many fond memories of play-by-email games of that...

In the late 80s, a friend of mine broke up with his girlfriend via email while they were both sitting in the same computer lab.