polynomial

10 hari yang laludiscuss

If a conscious model is born here, it should automatically be a (US) citizen.

12 hari yang laludiscuss

The most concerning thing here is the temperature problem. If your harness isn't providing deterministic output at a temperature setting of 0.0, it is broken.

polynomial

22 hari yang laludiscuss

You don't know that No Feigning Surprise is actually from an xkcd comic, before it was a wizardzines post? U+1F632

polynomial

22 hari yang laludiscuss

Honestly, if you compare Tim's pre-GPT writing style, to his current stylometrics, you'll see a remarkably delta. It's as if he suddenly decided to change his established writing style and voice right when LLMs become widely available.

What a coincidence!

polynomial

23 hari yang laludiscuss

Reasoning chains as first-class OTel spans, which is what makes autonomous agent production deployments auditable (rather than indefensible.)

polynomial

2 bulan yang laludiscuss

Less MTTF and more MTTR

polynomial

2 bulan yang laludiscuss

This is horrifying on so many levels even if you don't already know how the story ends.

polynomial

2 bulan yang laludiscuss

Calling it "Act 2" certainly isn't helping with investors.

polynomial

2 bulan yang laludiscuss

A 67 year old woman living out of her car? JFC.

polynomial

2 bulan yang laludiscuss

Very obviously LLM written to anyone who's spent any time using them.

polynomial

2 bulan yang laludiscuss

Not everybody isn't worried: https://ctolunchnyc.substack.com/p/cto-lunch-nyc-spring-2026...

polynomial

3 bulan yang laludiscuss

"The gap is larger than most people realize" is a a dead giveaway.

polynomial

4 bulan yang laludiscuss

Fully agree.

If you don't change your approach but just use CLI "intead of" MCP, you'll end up with a new spin on the same problems. The guardrails MCP provides (identity, entitlement, multi-principal trust boundaries) still need to exist somewhere.

https://forestmars.substack.com/p/twilight-of-the-mcp-idols

polynomial

4 bulan yang laludiscuss

This is the right framing. The chain policy problem is what happens when you ask the registry to be the entitlement layer.

Here's a longer piece on why the trust boundary has to live at the runtime level, not the interface level, and what that means for MCP's actual job: https://forestmars.substack.com/p/twilight-of-the-mcp-idols

polynomial

4 bulan yang laludiscuss

Understandable as always a proxy for predictable.

polynomial

4 bulan yang laludiscuss

This is the under acknowledged "secret" of this reconfiguration.

It's like the Bill Joy point about mediocre technology taken to the next level.

polynomial

4 bulan yang laludiscuss

Robots making fun of us complaining about them.

polynomial

4 bulan yang laludiscuss

Well at least you know who to fire