alexjurkiewicz

The Web of Trust failed for PGP 30 years ago. Why will it work here?

For a single organisation, a list of vouched users sounds great. GitHub permissions already support this.

My concern is with the "web" part. Once you have orgs trusting the vouch lists of other orgs, you end up with the classic problems of decentralised trust:

1. The level of trust is only as high as the lax-est person in your network 2. Nobody is particularly interested in vetting new users 3. Updating trust rarely happens

There _is_ a problem with AI Slop overrunning public repositories. But WoT has failed once, we don't need to try it again.

His latest post is endorsing a crypto exchange because they paid him $50k.

https://steve-yegge.medium.com/bags-and-the-creator-economy-...

Developer goodwill. And it probably cost a song.

The blog post says, many times, not to use Gastown. It makes fun of the tool's inconsistent branding and describes a lot of jankiness.

This tool is dangerous, largely untested, and yet may be of interest if you are already doing similar things in production.

Simply combine this tool with `openssl enc` and your shell script is as secure as any shell script could be

(2024)

My favourite part of these tools is the zany use of numbered file descriptors. `keypair` outputs the public key on fd 5 and secret key on fd 9. But signing reads the secret key on fd 8, while verification reads the public key on fd 4! Why aren't they the same?? I have to read the manpage every time.

I play several sports across several teams and leagues. Each league has their own system for delivering fixtures. Each team has its own system of communication.

What I want is software that can glue these things together. Each week, announce the fixture and poll the team to see who will play.

So far, the complete fragmentation of all these markets (fixtures, chat) has made software solutions uneconomic. Any solution's sales market is necessarily limited to a small handful of teams, and will quickly become outdated as fixtures move and teams evolve.

I'm hopeful AI will let software solve problems like this, where disposable code is exactly what's needed.

The article isn't talking talking about "industrial" in relation to user interfaces. It isn't talking about user interfaces at all.

Your consumer/enterprise/industrial framework is orthogonal to the articles focus: how AI is massively reducing the cost of software.

The "industrialisation" concept is an analogy to emphasize how the costs of production are plummeting. Don't get hung up pointing out how one aspect of software doesn't match the analogy.

I think part of this is that Status Page updates require AWS engineers to post them. In the smaller Tokyo (ap-northeast-1) region, we've had several outages which didn't appear on the status page.

You are absolutely write! Let me write a Go program to implement this idea. The bloom filters will take approximately 5gb (for a 1% error rate) and take a few minutes to populate on a modern Macbook Pro.

https://gist.github.com/alexjurkiewicz/1abf05f16fd98aabf380c...

ZeroTier supports relaying natively. You create a network with three nodes, the "client" (on the internet), the "gateway" (public subnet) and the "server" (private subnet). ZeroTier will automatically route traffic between the client and server through your gateway with no configuration.

I mean, LLMs have the same sorts of problem with

"Which poker hand is better: 7S8C or 2SJH"

as

"What is 77 + 19"?

It doesn't seem like the design of this experiment allows AIs to evolve novel strategy over time. I wonder if poker-as-text is similar to maths -- LLMs are unable to reason about the underlying reality.

S3 uses 5-of-9 erasure coding[1]. That's roughly 2x overhead.

[1] https://bigdatastream.substack.com/p/how-aws-s3-scales-with-...

It's hard to take this article seriously. We should write software for DOS because we won't need to maintain it post-release?

Maybe software written in the age of DOS was relatively trivial compared to modern tools. Maybe there's a benefit to writing code in Rust rather than C89.

> So am I saying don’t use LLMs? No, I’m not. I’m saying it needs serious guardrails, documentation, oversight, usage data, research, and it should definitely be banned from schools. Because it’s doing serious damage to young people’s ability to think, figure out, read, understand, and research.

The original article didn't talk about any of these things....

> My son is in school. And the teacher literally said they need to use an LLM otherwise they won’t score high enough because the other pupils are using LLMs and the teacher is using an LLM to grade/check it. This is ridiculous.

Ah. The article's author heard a bad take on LLMs and accidentally attributed it to Ptacek's article.

Yes. It's a small DSL to fork a repository and apply a series of textfile transformations (replace file, replace partial file).

But if you give this a cool name, it's a New Idea.

You can de/compress multiple streams in parallel, at minimal cost of overall efficiency. For gzip, look at pigz. For video, look at how x264/x265 slice up the video frame.

Right, this feels like an entry to Raymond Chen's "it rather involved being on the other side of this airtight hatchway"!

I have nothing against Signal implementing screenshot hiding. But it's not exactly fixing a gaping security hole.

I think code retention charts will look similar for any major library in any language. Projects accrete code.

You could instead consider:

* How many major version releases / rewrites happen in this language? (This might be a sign of ecosystem instability.)

* How much new code is replacing old code? (This might imply the language needs more bugfixes.)

There's nothing about ZeroTier's solution which deserves a higher price point than Tailscale. As a long-time user, ZT's administration UI is much worse and the product has been essentially unchanged for a decade.

Better solutions would be things that make the VPN invisible, rather than 'easy'. Tools such as Teleport, IOW.

Copilot was the first mover, but other tools (built on worse models!) can do just as well.

Until one product becomes clearly better, the code AI ecosystem might just churn and churn and churn.

> The only point of the program is to print the output.

A consistent structure helps humans parse the output. I have host(1)'s output order in my mental memory.

Context https://www.silverliningsinfo.com/automation/weaveworks-unra...

ProxySQL is a complex jet engine requiring manual tuning. 99% of MySQL users will be satisfied with the simple configuration and automatic query parsing of MySQL Router.

Corian isn't engineered stone under the Australian definition. It's explicitly mentioned as a safe(r) alternative.

Now they can sell advertising to businesses and guarantee improved CTR.

What prevents teams extending the expiry date repeatedly?

If you are reviewing implementation semantics after someone has already coded a feature, I'd say it's too late. You might catch issues, but the loss of goodwill and wasted productivity will make everyone hate code review.

I think this pyramid is accurate but better framed as a "review pyramid". Catch semantic issues in an earlier phase (informal discussion or rfc-style proposal document).