cortesi

I only had time to use it for a couple of deep reviews of large Rust projects, and a few agentic coding tasks (implement plan X, refactor Y in fashion Z) before my quota ran out. My impression is that the reviews were quite strong - maybe Opus 4.8+ or around GPT 5.5 (for my particular use case) - but very slow. For implementation I found it weaker, it made a few mistakes that I haven't seen frontier models make in a long time.

As a developer outside the US I think it's vital to have alternatives to OpenAI and Anthropic, but sadly this is not it. For $200/month you get < 3 hours of use per week, the API is extremely slow, and the output quality in my tests is nowhere near Fable. It's nowhere remotely near usable as a day-to-day workhorse. Very disappointing.

https://x.com/cortesi/status/2068898694238486658

I'm sorry, but we can tell absolutely nothing about Gary Marcus from this. People should have a look at the final data:

https://github.com/davegoldblatt/marcus-claims-dataset/blob/...

Many of the "supported" claims here are vague, banal, obvious, or just opinion. E.g.

"the general public hasn't quite realized what's not possible yet"

"loads of things scale, but not at all"

"To be sentient is to be aware of yourself in the world; LaMDA simply isn't."

"To date, nobody, ever, has given a convincing and thorough account of how human children (and human children alone) learn language."

"A cat holding a remote control shouldn't have a human hand."

"What I didn't see last night was vision" (about Tesla Optimus)

Written in Rust, with cross-platform + web visualiser in egui.

I have a tiny, relevant weekend project:

https://github.com/cortesi/agentsmd

This is a command-line tool that lets you generate your AGENTS.md and CLAUDE.md files from common sources. So, for instance, if you have Rust-specific guidance for models, you can define it once, and then automatically include it in any project that contains Rust based on the `lang()` language matcher.

This is one of those small tools I now use many times a day to maintain and update ubiquitous agents files. Maybe other folks will find it useful too.

SQLx is great, but I had a long laundry list of issues with its SQLite support so I forked it into a focused SQLite-specific library. It has now diverged very far from SQLx, and the number of small inaccuracies and issues we fixed in the low-level SQLite bindings is well into the dozens. The library is unannounced, but is already being used in some high-throughput scenarios.

https://github.com/cortesi/musq

Nextest is one of the very small handful of tools I use dozens or hundreds of times a day. Parallelism can reduce test suite execution time significantly, depending on your project, and has saved me cumulative days of my life. The output is nicer, test filtering is nicer, leak detection is great, and the developer is friendly and responsive. Thanks sunshowers!

The one thing we've had to be aware of is that the execution model means there can sometimes be differences in behaviour between nextest and cargo test. Very occasionally there are tests that fail in cargo test but succeed in nextest due to better isolation. In practice this just means that we run cargo test in CI.

I'd love to say that it's taking off, but it looks like daily active users (as measured by sampling popular relays) has been stagnant at just under 20k for 6 months:

https://stats.nostr.band/

We haven't found this to be an impediment. Keep things modular, and share the type definitions of anything you import with the model. As the benefits here become more and more clear tooling will improve and people will adapt their development practices to get the most out of the models.

Claude 3.5 Sonnet's coding abilities are incredibly impressive. I think it lets an expert programmer move more than twice as fast. There are limits - to produce high quality code, not copy-and-paste pablum, you have to be able to give detailed step-by-step directions and critically evaluate the results. This means you can't produce code better than you would have written by yourself, you can only do it much faster.

As an experiment, I produced a set of bindings to Anthropic's API pair-programming with Claude. The project is of pretty good quality, and includes advanced features like streaming and type-safe definitions of tools. More than 95% of the code and docs was written by Claude, under close direction from me. The project is here:

https://github.com/cortesi/misanthropy

And I've shared part of the conversation that produced it in a video here:

https://twitter.com/cortesi/status/1806135130446307340

Oh, right! That's just a bit of fun - a random pattern (picked to be vaguely visually pleasing), which is laid out on the Hilbert curve, and then offset by a calculated amount depending on the mouseover. So I'm just shifting the same pattern forwards and backwards in a ring buffer, basically.

The mousover always shows a contiguous section of the underlying file. So you're seeing a constant-length section of the space-filling curve centered on the cursor as you move the mouse about. The hex to the right is that same contiguous piece of the file. Sort of neat, actually.

There is, but I see it's not visible on smaller resolutions - I should fix that. The default color scheme just classifies bytes into black (0x00), white (0xff), blue (ascii), and low (green) and high (red).

Ah, this is an old side-project of mine. Something I should probably make clearer is that files are not uploaded anywhere - the app is completely local, and all analysis is done in the browser.

This version is written in React but when time permits I plan to release an updated version written in Rust, along with a library of fast implementations of space-filling curves and related utilities.

This was a fascinating rabbit hole. Here's Myron May - who shot and injured 3 people and then was killed by the cops - describing his "stalking". At first blush, he seems intelligent and rational - but as you listen, it soon becomes clear that he's in profound psychological distress.

https://www.youtube.com/watch?v=N-8PklxYM3w

He's using all the terminology employed by the targeted individual community - he clearly found "his people" online, and with it validation and an explanation of what he was experiencing. I wonder what role this played in the fact that he never got the help he needed.

I once correctly guessed a that friend's PIN was "1729", based only on the fact that he was a maths major, a huge fan of Ramanujan, and was sure to have read this story. I still cherish the look of complete confusion on his face, more than 20 years later.

It depends on what kind of geek you are. "The Ants" is perhaps my favourite non-fiction book of all time - at the time it was written it was a fair account of the total sum of human knowledge of ants. It's beautifully written and incredibly complete, and although it's now a bit dated, it's still the best reference to ants if you're interested in them. For a broader summary of Wilson's intellectual project I recommend "Consilience" - a wonderful synthesis of almost all his ethical and scientific interests.

This review of "Naturalist" in the NYTimes gives some flavor of events:

https://archive.nytimes.com/www.nytimes.com/books/98/12/06/s...

The book itself is well worth reading - it's a wonderful scientific autobiography written by, for my money, the best science writer of the last century.

The feud is alluded to in the article. The broad summary is that Wilson had the temerity to devote the final chapter of Sociobiology, a book that pretty much launched the field of evolutionary study of animal behaviour, to humans. Lewontin and Gould - among others - had ideological objections to this, resulting in a vicious personal persecution of Wilson that stretched over decades. This is all very well documented, but to appreciate the sheer unfairness of the way Lewontin prosecuted his vendetta you have to hunt down Wilson's accounts of the affair in books like "Naturalist". It's very hard to come away from it without having your opinion of both Lewontin and Gould soured.

Lewontin's reputation will be forever tarnished by his disgraceful treatment of E.O. Wilson, a dispute in which he was viciously unfair and in which Wilson has proved to be correct in every particular.

That's quite possible! Though after staring at thousands of these across many application domains, I do think the space-filling curves perform legitimately better.

Another factor is that the large discontinuities in the zig-zag curve means that a contiguous area in the data is not always contiguous in the visualisation, which makes things like the region selection I do for https://binvis.io impossible.

Oh, hey - that's my post. I have a blazingly fast re-implementation of these algorithms in Rust, which will soon make an appearance as a set of command-line tools. This will be a companion to the interactive in-browser implementation at https://binvis.io.

If this sort of thing floats your boat, please get in touch. The code is not public yet because it's not ready yet, but I'd welcome like-minded collaborators.

Whoa. Many, many years ago I started idly hacking on a similar project with similar ideas and an identical name. I've always intended to return to it in my Copious Spare Time, which of course means I never did and probably never will.

When I started exploring this I registered the gitbug.org domain - would you be interested in using it for your project?

Taleb's prominence is baffling to me. I've read all his books, and tried hard to figure out why people I respect believe he's an important thinker. I just can't see it. Almost everything he says sets off my bullshit alarm.

Here we have him in typical form. The point of the paper seems to be to argue that we can reduce transmission by reducing contact (obvious), and concludes that we should do this pre-emptively at large scale worldwide (never going to happen, only good for alarmist headlines). This is exactly his formula when treating risk: start with a total platitude (rare events happen), and spin it out through huge over-reach into a headline-grabbing book (black swan! boo!).

The other thing about Taleb is that he's frequently way out over his skis on on the facts. Here, he states that the "selective dominance of increasingly worse pathogens" makes "extinction certain" as if he's stating an undisputed fact. But instead of citing a virology paper, he cites an interesting but not very relevant evolutionary dynamics paper that draws on a particular mathematical model. In fact, the real world is complicated, and if anything virology tells us the opposite: zoonotic viruses tend to become less virulent over time, as the tradeoff between transmission and lethality is optimised. Almost nobody working in the area of virology would agree with the alarmist nonsense that this paper takes as axiomatic.

If anyone is interested in hearing how actual virologists talk about this outbreak, the superb This Week In Virology podcast has just released an episode that dives into this in depth:

http://www.microbe.tv/twiv/

Excellent. I did something similar a few months ago using a dataset of zoological silhouettes, resulting in a menagerie of mammals, bugs, spiders and other mutant wonders.

https://twitter.com/cortesi/status/1153075801891278848

https://twitter.com/cortesi/status/1153088629972934656

This matches my experience with BigTable, down to the short-duration failure spikes.

I feel that something should be said on the plus side of the ledger here. I'm the solo founder of a company that indexes huge amounts of fine-grained information. Bigtable is the key technology that let me start my company on my own: it soaks up all the data we can throw at it, with almost zero maintenance. Even within the stable of GCP technologies it stands out as being particularly reliable.

My biggest "problem" with BigTable is the lack of public information on schema design - which in this context is mostly the art of designing key structures to solve specific problems. I've come up with sensible strategies, but much of it was far from obvious. I can't help but feel that there should be a body of prior art I could draw on.

I find the rank graphs in this post a bit unsatisfying - I keep wanting to see what the relative magnitudes are. I've been experimenting with a visualisation that combines rank and magnitude. It needs another few iterations but I think the direction is promising:

https://netograph.io/data/stats/connections?q=all

Mr Puzzle did a set of videos exploring the flaws of the ISIS Orb:

https://www.youtube.com/watch?v=R0sCGotCw2Y

The Mr Puzzle channel as a whole is highly recommended if you're into puzzles of this sort.

I have a project that's relevant here - netograph.io captures low-level data on website behaviour, then indexes the data in various ways for querying. Right now, it ingests a sizeable fraction of links on social media live. Here's my data for the api.behavioraldata.com domain, which is the first tracker they mention:

https://netograph.io/datasets/social/domain/api.behavioralen...

It's interesting that Netograph has seen this exclusively on .pl domains, and that it hasn't cropped up again in the last month. You can do similar digging through the dataset for all the other trackers they list.

Building more understanding of file formats into binvis is perpetually on my todo list. This would let you interpret different segments of files differently, and correctly interpret instructions and other higher-level structures. It would also help with things like entropy calculation over sliding windows (where you don't want this to span file segments). There are so many interesting avenues to explore here.

While I'm thinking about it, you're of course quite right that spotting periodic patterns in a Hilbert curve layout is really hard. However, I've found that being able to spot them in a scanning layout is highly dependent on your column width matching/resonating with the period of the pattern you're looking for. Another thing on my todo list is to combine Hilbert layouts with Fourier transforms. There's a very suggestive paper on this that I'd love to explore:

https://arxiv.org/pdf/1508.03817.pdf