"Academic" "computer" "scientist" but really none of those things. I program, think, write, talk, teach... rarely in that order. Mostly I make software slower.

Joined September 2011
18 Photos and videos
Stephen Kell @stephenrkell@recurse.social retweeted
To: the crazy ones, the misfits, the rebels, the troublemakers, the ones who see things differently Subject: you should write an Onward! paper alarmingdevelopment.org/?p=1…
3
32
79
17,077
I'm going 'on tour' soon... among others, going to be in the Bay Area on Tuesday 7th and Wednesday 8th November. Happy to chat with anyone who's around! Also will have some time in New Zealand and Australia the fortnight after that...
4
672
No coincidence that the first such statement comes from one of few institutions retaining a shred of academic self-governance. (But no time for Cam-complacency. The votes were alarmingly close, no doubt owing partly to recent Regent House gerrymandering.)
"Cambridge Uni and UCU call for an end to marking boycott in joint statement" thetab.com/uk/cambridge/2023…
1
993
Saw a tweet (or was it a toot?) in the last week or so about how apparent AI "advances" are at least partly a function of giant compute resources being owned by few companies -- clearly true. Like a fool, I did not 'like' it. Can anyone help me find it? Looking to mine this seam
3
706
Does anyone know a good reference for static analysis taxonomy like "flow-sensitive", "context-sensitive" etc.? I think I absorbed them as folklore, but hopefully they're described somewhere. Some years ago I took on a related question (static vs dynamic) humprog.org/~stephen/blog/…

2
8
1,485
Interesting post, interesting discussion but (my take) missing the elephant. Languages grow huge because existing approaches to interop are so dire. This amplifies the incentive to pile all features into one language, rather than express parts of a codebase in languages that suit
How Big Should a Programming Language Be? tratt.net/laurie/blog/2023/h… 1/
3
12
2,269
Popped into the new West Cambridge library (sorry, "high-intensity study space" -- you can't make it up) and chanced upon some academic texts that seemed a bit too appropriate.
2
576
Although it's provocative to label bikes as 'unsustainable', there are lots of interesting observations here about modern manufacturing, materials, and market dynamics all conspiring to worsen long-term value. solar.lowtechmagazine.com/20…

1
3
783
Keeping Microsoft Teams running has become a daily permacrisis. MS seem to have stopped offering the Linux client (empty dir?!); web one doesn't work in Firefox; Chrome now randomly crashes with GPU-related errors; Android one is now flaky on my old phone. packages.microsoft.com/repos…

1
3
5
800
A razor-sharp thread well worth reading, not just for the references to "stochastic parrots" and "Hapsburg AI". Declining signal-to-noise is the tragedy of our age, quality being replaced by superficial plausibility... yet S/N was Google's initial strength.
The really remarkable thing isn't just that #Microsoft has decided that the future of #search isn't links to relevant materials, but instead lengthy, florid paragraphs written by a #chatbot who happens to be a habitual liar - even more remarkable is that #Google agrees. 1/
1
4
908
Just joined Mastodon by following these steps. They rely on an API due to get paywalled off *tomorrow*! fedifinder.glitch.me/ Also, an additional essential step (for me): switch to "light mode". I really don't know how people live in the dark like that without falling asleep.

1
2
488
It seems common that university courses on operating systems include a bit about how to implement malloc() and free(). But why? It seems out-of-place, needing so much awkward caveating: "we are now [potentially] in user-space"; "unrelated to all that page-table stuff"; etc.
6
3
2,747
Generalised Thing Bothering Me today about academia: functions taken away from academics in the name of "support" and "efficiency", but being done badly, causing externalised fallout on those nominally being "helped". Web pages and other informative stuff are a recurring example.
2
15
1,644
(This tweet hasn't aged well... by lunchtime there was a whole other Generalised Thing Bothering Me that is way more annoying, although probably too big for a tweet. But the theme is the same: the corporate-style university has neither means nor incentive to preserve quality.)
1
5
437
This article is interesting. But I would love to see an analysis that splits issues "inherent to any similar service" from "accidents of Mastodon's design" or "bad/misinterpreted regulations".
26 Jan 2023
This is an excellent piece. The FT learns first hand why tech compliance officers and sysadmins have such gruelling jobs. on.ft.com/3XzJ8Sv
1
1
702
My inclination is always to compare to Usenet days. Certainly there was ISP overhead, but it wasn't crippling, despite huge volumes of porn/piracy/spam. Is the difference just by modern-day scale ('everyone is online'), or that NNTP built in better {scale, deniability, ...}?
1
316
Until today I thought "statistical multiplexing" was an idea basically every computer scientist would know... ... but I notice it doesn't have a Wikipedia page, and the top search results are super-specific to networking, which the idea is not. How giant is my misapprehension?
3
4
1,065
I spend far too much of my life fighting presentation/diagramming software that doesn't work. Does anyone have a solution that isn't either "I use [Windows/Mac]" or "I mastered Tikz and then wrote a custom UI layer for Inkscape"?
12
4
29
7,015
A question for people who know about SMT.... If I have a formula over some variables x, y and z, then I additionally suppose x and y have unique (but unknown) values, are there any good ways to reason about when z also has a unique value?
1
706
It's suspiciously like solving school-style equations (but maybe doesn't need to be only about equality?). Or something about extracting functional dependencies that are latent within a formula, or treating formulae algebraically, or...
315