dools

The #1 goal for Anthropic and others is to take the longest running process possible and make it entirely opaque to the developer. It's the only way they can build a moat for a commodity. I would highly recommend building your own multi-stage orchestration flows because then you'll get a much better idea of where you need to be in the loop, and where you can save money. Once entire organisations are functioning only as extensions of Anthropic, they'll put the prices up and squeeze the shit out of the market.

I've bootstrapped a telecommunications product[1] to low 6 figures in ARR. It's the most successful venture I've had. I started the very first version in 2018, and released the first version of the app in 2021 (before that it was entirely done using off-the-shelf tools). I'm mid-40s now, and have been either freelancing or trying to launch products since I was mid-20s. I still have a job, but it's a good one that I enjoy doing and it leaves time for my business stuff. I would be better off financially if I had just focused on a traditional engineering career path and invested regularly in broad based, index fund ETFs after finishing university in my mid-20s, but whatever. I own a house in Australia which is an achievement on its own. I have a lovely wife and 2 teenage kids. We're all going on a 2 week holiday to Greece later this year. I have 2 new cars (financed) and a nice office in the city. I have always been the primary income earner for my family of four. Life gets more expensive as my income goes up so I don't wind up with much left over even though I'm earning considerably more now than I ever have. I'm pretty happy with how things turned out and I've always enjoyed the various challenges I have attempted, but on the other hand I look like I'm about 60 and I've lost a lot of sleep!

[1]https://www.benkophone.com/

I think they’re referring to services using AI for coding and shipping bugs more often as a result.

> Does U.S. intelligence know something we don't?

Well, yes probably.

“ None of this bites evenly. The editing falls hardest on broadcast and promotional push; the notifications people actually want tend to pass through untouched or amplified”

So … mission accomplished then? This is pretty much how I would like it to operate.

The funny thing is that as much as everyone complains about AI overviews and Google losing the plot and so on, about a year and a half ago the “where did you hear about us” field on my website stopped being dominated by “Google” and started being dominated by “ChatGPT”.

Normal people are using AI for search more already, Google is just trying to stop their primary business from completely disappearing.

Sounds like your employer would definitely be better off paying you to stay home …

Don’t let the door hit you on the way out!

Yeah, prediction markets are like if you could bet on arbitrary decisions a coach or manager might make. Like starting lineup or different trades of players between teams or the size of a players contract in dollars, rather than events that happen during gameplay. There are people who know the facts already but the public does not.

> They ingest a lot of source code, hold a single hypothesis at a time, and iterate against it. That's exactly the wrong shape for vulnerability research, which is narrow and parallel by nature

LLMs are trained on Ed Sheeran lyrics

I think creepiness manifests when the observation is one way. Without technology that’s kind of hard. With tech it becomes increasingly easy for everyday people to one-way spy on each other

This was one of my mother’s favourite shows and we all watched it together as a family when I was a kid. I really enjoyed it too.

About 10 years ago I bought the DVDs for my mother for her birthday and we watched some together, it really stood the test of time if you ask me.

She died last year, I’m glad we got the chance to share this and a few other nostalgic things over the past few years, we lived quite a long way away.

Call your mother!

I said "open weight" rather than "local". I mean, local if you have $240k to drop on GPUs but you can run Kimi k2.6 on a B300 cluster for ~$50/hour too.

Kimi k2.6 is about on par with GPT 5.2 so I’d say open weight models are about 6 months behind.

I’ve noticed that the emphasis in messaging and product from Anthropic is towards monolithic agent usage rather than building systems using agents or building more specialised agents. I listened to a talk by Boris recently and his vision for the future was that “the model just knows”.

My guess is that they are trying to increase the cost of switching as much as they possibly can before the VC subsidies run out and they have to 10x their prices.

I’ve been using kimi 2.5/2.6 for the past 2 weeks and it’s really not far off OpenAI and Claude models. I am a coder so it’s not all vibes but I am definitely more in the “spec to code” mode than “edit this file for me” and it copes just fine. Needs a bit more supervision than the frontier models but it’s also significantly cheaper. If I were anthropic I’d be shitting myself, their prices are going to 10x over the next 2 years

Verda has B300 clusters, 8 for USD $55/hour in 10 minute billing blocks

Check out Verda you can rent whatever super powerful GPU clusters you need in 10 minute increments. Deploy any open weight model using SGLang and away you go

This is the most compelling "who is Satoshi?" post I've found:

https://bitcointalk.org/index.php?topic=628344.msg48198887#m...

"I contend that James Simons put the team together that made up Satoshi Nakamoto and that Nick Szabo was the main public-facing voice behind the nym."

Same, I tested Claude Code CLI and it crapped out and took my money, and they haven't responded to my billing dispute yet. Meanwhile JetBrains replies within hours and Junie is LLM agnostic. I'm a huge fan of JetBrains for AI coding.

No editorialising in the title

Uh oh did someone CVE grok or twitter?

The schema describes the database layout. The file layout (if you were going to call it that) in a modern RDBMS would describe how the RDBMS implemented a particular database layout as described by the schema.

Trump is a liar, he lies about everything. Starting a sentence with “trump said” implies that any concurrence with reality is purely coincidental.

Easy: this is political retribution.

USAID: 0.6% of the budget CFPB: 0.011% of the budget

It has nothing to do with saving money and is well beyond the executive order than instantiated the agency.

If all they do is disrupt things enough for some crypto dorks and Russians to make a play it was all worth it.

Wrestling

Given the fact that there are so many Musk apologists and "anti-woke" crusaders in the HN community, most notably Paul Graham, they probably won't do it.

HN has a pretty "IBM During Nazi Germany" vibe.

Don't forget all the zombie drones who attacked the capitol on his behalf

This pardon is corrupt. Ross' parents donated to Trump and he pardoned their son as a favour.

Whether or not you think he deserved the prison time, the problem here is how utterly brazen Trump is in accepting bribes.

An amazing example of a DSL was Butler for Trello before it was acquired and dumbed down to a “no code” tool. It was magical to use. I had started consulting using it but they said I wasn’t allowed to use it anymore so I replicated the functionality as a library to run inside a Google sheet using Google apps script.