jacobgold

472 karmajoined 5 tahun yang lalu

Hello friend, nice to meet you.

I'm also on Bluesky: https://bsky.app/profile/jacob.gold

Always happy to chat.

[email protected]

Submissions

[untitled]

1 pointsby jacobgold3 hari yang lalu0 comments

[untitled]

1 pointsby jacobgold8 hari yang lalu0 comments

Show HN: Coding Agent Survey – Which coding agents do you use?

codingagentsurvey.org

5 pointsby jacobgold9 hari yang lalu4 comments

[untitled]

1 pointsby jacobgold10 hari yang lalu0 comments

[untitled]

1 pointsby jacobgold11 hari yang lalu0 comments

[untitled]

1 pointsby jacobgold25 hari yang lalu0 comments

Anthropic's new Agent SDK pricing is a win for Codex

4 pointsby jacobgold26 hari yang lalu1 comments

CLAW.md – open format for agentic cron jobs

5 pointsby jacobgoldbulan lalu1 comments

Show HN: Clor – give your agent claws

11 pointsby jacobgoldbulan lalu5 comments

Stop Telling Users Their DNS Is Wrong

5 pointsby jacobgold5 bulan yang lalu1 comments

comments

kemarin duludiscuss

Maybe Zuck should double down on his "spoiler" role with models rather than compete head-to-head.

He doesn't have to match Anthropic or OpenAI model revenue if he can deflate theirs by 99%.

All he has to do is keep spending a few billion dollars developing frontier models, release them as open weights, and turn coding models into a commodity. He also needs a good OSS reference harness to match. Very few people are in a position to do this and for it to make business sense.

That's quite likely where things are headed regardless, and he could speed it up significantly.

We should all hope models move from proprietary products to commodities the way compilers did.

This may be one of the best things Zuck could do for the world.

kemarin duludiscuss

This project looks great and I wish them luck. I'm just lamenting the fact that we still haven't solved the really big problem with Slack-like chat apps.

The Tailscale analogy isn't quite right because there are no real network effects involved. Most of Tailscale's utility exists even if no one else uses it.

Slack is only useful if your friends, coworkers, or partners use it. Same with Discord, and even open source alternatives for the most part.

kemarin duludiscuss

You want the open protocol to have network effects, not a proprietary company's product.

Email worked out pretty well, while IRC failed for reasons that are probably correctable.

kemarin duludiscuss

The fundamental problem with replacing Slack is network effects. Your coworkers and customers already use Slack. It works well enough.

You can choose to switch your company away, maybe, but what do you do when vendors want to connect over Slack?

Imagine if email was owned by a company?

Edit:

W̶e̶ ̶r̶e̶a̶l̶l̶y̶ ̶n̶e̶e̶d̶ ̶a̶n̶ ̶o̶p̶e̶n̶ ̶p̶r̶o̶t̶o̶c̶o̶l̶ ̶t̶o̶ ̶b̶u̶i̶l̶d̶ ̶o̶n̶.̶

We really need an open protocol to win here.

kemarin duludiscuss

> "The best feature is that it can delegate questions out to GPT-5.5 in the background, so you're no longer restricted to a voice model that's several years behind the frontier."

Wow. That's exactly what I hoped they would do.

This issue has held me back from using ChatGPT's voice mode as much as I otherwise would have, because I also use it for brainstorming while commuting, exercising, etc., and don't want it to feel stuck in the past.

3 hari yang laludiscuss

This is a paper on ecotoxicology and environmental safety, so describing aluminum as an "environmental contaminant" makes sense in this context.

> "Did you know Dihydrogen Monoxide is found in 100% of cancers?"

Did you know humans intentionally consume water because they need it to survive, and do not intentionally consume aluminum?

4 hari yang laludiscuss

> "There's no risk to leeching, there's no data proving any link at all between Aluminum and Alzheimer's."

EFSA (European Food Safety Authority) says acidic or salty foods can increase aluminum leaching from pans and foil into food: https://efsa.onlinelibrary.wiley.com/doi/pdf/10.2903/j.efsa....

This review found mixed evidence and a positive meta-analysis signal for aluminum exposure and Alzheimer's: https://www.sciencedirect.com/science/article/pii/S014765132...

It's not that hard to avoid this leaching concern, seems smart to make a basic effort to do so.

4 hari yang laludiscuss

Like most metals, excess aluminum in your diet is almost certainly bad for you.

EFSA full review: animal studies show nervous-system effects from aluminum exposure. https://efsa.onlinelibrary.wiley.com/doi/pdf/10.2903/j.efsa....

CDC/ATSDR: oral risk levels are based on neurological effects seen in exposed animals. https://wwwn.cdc.gov/TSP/ToxProfiles/ToxProfiles.aspx?id=191...

4 hari yang laludiscuss

You shouldn't use aluminum foil with tomato/lemon/vinegar/brined foods or as "grilling packets" (direct fire/high heat) because of leaching risk.

There's no concern with using aluminum in most cases (with dry/non-acidic foods) but leaching is a real problem with acidic/salty/wet/high eat.

5 hari yang laludiscuss

> "They are intended to trigger anxiety, which is manipulative and (in this case) precisely contradicts the concern the author is purporting to have for the rest of us."

What? That's a very uncharitable take.

I wrote this blog post and it was a sincere attempt to genuinely help people. It seems to have been very helpful for some people, based on the responses/messages I've received.

7 hari yang laludiscuss

Fable is supposed to return to subscription plans, unless I'm missing something: https://jacob.gold/posts/fable-5-removal-is-temporary/

  Anthropic says the change is about capacity and is temporary. In its launch announcement on June 9, 2026, it says:

  "After this point—when sufficient capacity allows us to do so—we aim to restore Fable 5 as a standard part of subscription plans. We intend to do this as quickly as we can."

7 hari yang laludiscuss

"Nobody goes there anymore. It’s too crowded."

No matter how interesting something is, once you get over the novelty, it will never be exciting in the same way something new is.

People are doing a million interesting things, but yeah, there are also a lot of people doing very similar things.

That's because we're still figuring out how things should work in this new agentic world.

7 hari yang laludiscuss

> "~$40k At this price level, you get the next step up in model intelligence. Something pretty close to Claude Opus."

That is equivalent to 16.8 years of Claude Opus 4.8 or Codex GPT 5.5 at $200/mo.

I'm a huge fan of running local models, but they're still wildly expensive, lower quality, and possibly dangerous (if backdoored). I sincerely wish this wasn't the case.

8 hari yang laludiscuss

> "Also if you are using local AI that you didn’t train yourself you can never be sure..."

A local model you trained yourself seems about as good as you can do today.

But it may not even be possible to fully trust a model you trained if you used untrusted data during training.

As a user, you have to trust your coding agent AND inference provider AND models: https://jacob.gold/posts/coding-models-are-code/ https://www.anthropic.com/research/sleeper-agents-training-d...

9 hari yang laludiscuss

The goal is some more real/accurate data we can look at month-by-month. People say a lot of things on X that aren't true.

9 hari yang laludiscuss

Things have changed so much over the last year. We now have terminal, desktop, mobile, and web agent UIs.

I'm trying to continually survey other devs so we have good data on what agents are actually in use.

10 hari yang laludiscuss

Most people don't care enough to argue at all. But no team ever created anything great without a lot of arguing. It's the only way to get to a "best idea wins" culture. It has to be productive arguing to be useful though, and it has to stay non-toxic to be sustainable.

Even on the best teams you should expect arguments to go off the rails sometimes. It takes real experience to learn how to argue well across a bunch of different personalities. When you get it right, arguing is genuinely fun and productive for everyone involved, and that's how you know you're doing it well.

10 hari yang laludiscuss

[dead]

10 hari yang laludiscuss

> Internet Archive? Non profit. Let's Encrypt? Non profit. ICANN? Non profit. Linux Foundation? Non profit.

Non-profits are great and we should have them too. If you look into how these non-profits are funded, it's largely corporate money.

10 hari yang laludiscuss

> "That also means the client itself deserves scrutiny. If a coding agent can read your repo and run commands, the binary that ships it should be boring (ƒor example, pi harness)"

You're actually trust your security to your harness AND model AND inference API provider in this scenario: https://jacob.gold/posts/why-i-wont-run-untrusted-models/