internet_points

Same here, I've always used em dashes and have been called out on negative comparisons – I didn't even know they were an LLM thing. Should I read more LLM to know what phraseology to avoid, or will doing that nudge me towards sounding more LLM? :-(

Oh, I thought it was in all of Philadelphia, but it's just inside courtrooms :(

All that on a single Github Sponsor[0].

[0] https://github.com/sponsors/Lakshmipathi

Yes! That one's going in my $PATH. Such a useful use of cat!

This regression towards the mean is still very much a feature of the newer models, in my experience. I don't see how a model that predicts the most likely word based on previous context + corpus data could possibly not have some bias towards non-novelty / banality.

> You have to go read it yourself afterwards

^ this is important.

Otherwise you may very well be missing anything really surprising or novel.

See for example https://www.programmablemutter.com/p/after-software-eats-the... , an experience report of NotebookLM where

> It was remarkable to see how many errors could be stuffed into 5 minutes of vacuous conversation. What was even more striking was that the errors systematically pointed in a particular direction. In every instance, the model took an argument that was at least notionally surprising, and yanked it hard in the direction of banality.

Also tokens don't represent meaning in themselves, but are assigned points in a multidimensional space, they can only represent meaning in the network as a whole when combined with other tokens in context and order.

And the abstract concepts of Abstract Wikipedia are human-defined, top-down ways of carving the world into distinct categories which make some kind of logical sense, whereas LLM's work bottom-up and create overlapping, non-hierarchical, probabilistic networks of connections with nearly no imposed structure except the principle that you shall know a token by the company it keeps.

But you can type them both out with keys on a keyboard so in that sense I guess they're not that different.

> Let's follow one example: Nigeria is the most populous country in Africa. In Abstract Wikipedia, this might be stored as: Z27243(Q1033, Q138758272, Q6256, Q15, Z27243K5)

Haha that's like John Wilkins' "Real Character, and a Philosophical Language"

https://en.wikipedia.org/wiki/La_Ricerca_della_Lingua_Perfet... is a great intro to the weird and wonderful world of abstract/universal/ideal/a priori languages.

> “Any distributed system based on exchanging data will be replaced by a system based on exchanging programs.”

So distributed systems tend to converge towards being more and more mystifying? Cf. the mythical mammoth:

> Show me your flowcharts and conceal your tables, and I shall continue to be mystified. Show me your tables, and I won’t usually need your flowcharts; they’ll be obvious.

If you read https://www-cs-faculty.stanford.edu/~knuth/papers/claude-cyc... it was more of a guided effort to write a program to find examples that helped with moving the proof along

yes

> This gets the binary down to 64M for me (a further -23%).

while on my system, /usr/bin/pandoc takes 199M; not bad.

I've tried figuring out what the big deal about cybernetics was, but I always come away with a feeling of it being a bit wish-washy. Is it a bit like Philosophy in that it birthed individual fields that were inspired by and made applications of the thoughts, models and ideas laid out by its forebears? Or were there actual proofs, discoveries or applications in the field itself?

(I guess one could call projects like https://en.wikipedia.org/wiki/Project_Cybersyn an "application" of its ideas, though cut off before one could see the results.)

See also recent post "A sufficiently detailed spec is code" which tried and failed to reproduce openai's spec results: https://hn.algolia.com/?q=https%3A%2F%2Fhaskellforall.com%2F...

The problem isn't getting an AI agent running in a sandbox. That's trivial. The problem is getting an existing enterprise project runnable inside the sandbox too, with no access to production keys or data or even test-db-that-is-actually-just-a-copy-of-prod, but with access to mock versions of all the various microservices and api's that the project depends on.

Well done for actually trying and failing to reproduce their results.

But the point from the original comic version kind of makes it seem like specs are useless, which I don't believe is true. Just like reimplementations can be useful (directly so for property testing or golden testing, but also for learning what things worked and what didn't), a specification can be useful even though it is informal or incomplete in that it guides the real implementation. (Like "plan mode", for you vibers.)

Making snark comments about that sounds very unlikely. More likely they'd have respect for someone living frugally and not showing off. See https://en.wikipedia.org/wiki/Law_of_Jante

> satire

I'm sure they've already received offers from investors who wish to build the next torment nexus.

Drawn in by the Emacs, learnt something new about C and C++, thank you for this! Very readable article for someone who doesn't feel too confident with low-level bits.

Btw, is this representation the reason why OCaml's ints are not as big as C ints?

Also interesting that the Haskell pointer tagging you link to[0] was done the way it was to avoid CPU branch misprediction, and that the old way which it replaced was "the source of half of the branch misprediction events". I wonder how "branch prediction friendly" current Haskell is.

[0] https://simonmar.github.io/bib/papers/ptr-tagging.pdf

I love Emacs, and while that tour doesn't quite look 1978, it sure doesn't look 2026 either. Maybe somewhere in the middle.. yup, 2002 seems about right.

And for some reason, most of the screenshots on that page are in super low res, e.g. I see https://www.gnu.org/software/emacs/tour/images/gnus-small.pn... instead of https://www.gnu.org/software/emacs/tour/images/gnus.png , I really don't see why they'd want to make it look even worse.

And then on the other hand you have people like Nicolas Rougier https://www.reddit.com/media?url=https%3A%2F%2Fi.redd.it%2Fp... or Álvaro Ramírez https://xenodium.com/awesome-emacs-on-macos who show how great it can really look.

I want this but for Haskell :-) Maybe some of the amazing work on dataframe[0] and related libraries could be used for a better Haskell REPL in Emacs. I never much liked the notebook way of working, I prefer having a file of functions alongside a REPL, but time-to-graph is bad, and I don't know if there's a good solution to how the REPL forgets previously defined variables on reloading a file.

[0] https://dataframe.readthedocs.io/en/latest/exploratory_data_...

This was a beautiful article; the joy of tinkering just shines through everywhere :-) I'm glad Rahul did the work to upstream some of the changes, I hope some of the maintainers read his post and are inspired to change a few defaults too (with that init.el vs `lisp/` refactor it should be easier for them to see what wants changing). Maybe some of the packages might provide for inspiration too; I'd love to see vc-mode provide builtin gutter support, for example. And viper extensions sounds like something that should just be upstreamed. (Less so exchange rates and weather.) Emacs is so close to being quite good out-of-the-box.

I use it that way. It requires less setup than lsp-mode. Just try it – add `:disabled t` to your use-package lsp-mode, restart emacs and type `M-x eglot` in a C++ buffer.

The Lem editor[0] and LispWorks IDE's[1] are implemented in Common Lisp.

Still, the reason for choosing a language for whatever are always more social and path-dependent than technical (reason 1: initial developer of whatever really likes the language, reason 2: language is seen as hip within some crowd, reason 3 (later in the game): management feels language is safe). Technical reasons for choosing a language typically tend to be post-hoc rationalizations. (I mean, no sane person would choose Javascript for an editor based on technical reasons alone, yet here we are.)

[0] https://lem-project.github.io/ [1] https://www.lispworks.com/products/lispworks.html

no android version though?

No, because of the legal principle of habeas pecuniam (you can't afford as many lawyers as Google)

and messaging etc. making it a social network

Well that's simply not true. There's lots of shared history ( https://en.wikipedia.org/wiki/Cognate ), common ancestor-words, and of course modern loans like "lacquer" and "crimson", but the "parents" of Sanskrit and Germanic languages diverged before Sanskrit appeared around 1500 BCE (earliest assumed date of Vedic Sanskrit, same time as Mycenaean Greek), there's lots of language trees online that make this clearer

https://jacklynch.net/language.html

https://learn.kids4alll.eu/sites/default/files/unit/oer/doc/...

https://www.theguardian.com/education/gallery/2015/jan/23/a-... is a beautiful one

Some of the animated adaptations were freakier than others: https://www.youtube.com/watch?v=aVQlJDtzUbo

(And even the books, floods and comets, children's books about impending natural disasters, not of the magical kind that you know aren't real anyway, but of the kind that have actually destroyed life on earth before and might happen again, that's real nightmare fuel for active children's imaginations.)

I have already bought Journelly for iPhone thinking it might be useful for such tracking, but the UI didn't quite fit. But I liked that it had an org-mode backend and no subscription/server part. So going by past data[0] I think you have at least one sure sale if you publish an iPhone version :-)

[0] https://www.momtestbook.com/