← Feed

Posts

• Post #4502684

  But good faith only applies to *moral agents*, and _morally_, (even if I were to concede a maximalist position as to their _efficacy_) an LLM is a random number generator. It might produce some malware to slip by you. You cannot "hold it accountable", there's nothing there to be held to account.

• Post #4476100

  RE: https://mastodon.social/@limebar/117060315235037705 I absolutely endorse this sentiment but I would be remiss if I did not observe that solidarity is, in addition to a general attitude, also a bunch of excruciatingly tedious *work*. You’ve gotta give up on your own priorities to make someone else’s more important. You’ve gotta figure out WHICH person you are going to do that for. You’ve gotta reach out, they’ve gotta respond, you’ve gotta hammer out a revshare agreement, you have to schedu...

• Post #4460248

  right-angle USB-C adapters technically violate the spec because they are extension cables and thus invalidate the capability detection of the e-marker chip, but that is a _super_ annoying technical detail and I'm glad they exist anyway

• Post #4396685

  I finally realized that there may be a more disturbing reason that every booster uses the exact same phrases, right down to the wording — "you'll be left behind", "it's here to stay", "it's a tool" — even when they are barely coherent and clearly out of context for the discussion at hand. That reason is hard to understand if you haven't played SUPERHOT though. You have to check out this game. It's the most innovative shooter I've played in ye...

• Post #4396216

  Have you ever worked in an organization where you were directly aware of how open source software (maintenance, devrel, upstream contributions, releasing internal code, operating an OSPO) was accounted for on the business's chart of accounts? I would love to hear from you about it.

• Post #4339603

  Really wondering if these Thought Leaders who are now — at this late moment, now that the KOSPI is tanking and Nadella is saying that trusting OpenAI will annihilate your business — are coming around to the idea that maybe LLMs aren’t really all *that* useful, will give *any* credit to the people who have been saying it for years

• Post #4290008

  RE: https://aus.social/@perkinsy/117006314371109767 “IN THE REAL WORLD, INDIVIDUAL PERFORMANCE IS RARELY THE STRONGEST PREDICTOR OF GROUP OUTCOMES. IN FACT, THE STRONGEST PREDICTORS OF DEVELOPERS' SELF-REPORTED PRODUCTIVITY TEND TO BE SOCIAL” (emphasis mine)

• Post #4261478

  you can order it now: https://keeb.io/products/sinc-lm-low-profile and I just did (oof.wav). if this is the kind of keyboard that appeals to you (split, staggered, mechanical, low-profile, EXPENSIVE), this is it.

• Post #4218384

  RE: https://mastodon.social/@glyph/115647478978972835 ah shit people are faving this again, what fresh hell awaits me when I find out why

• Post #4210041

  I need an invoicing solution. I want something that can send emails, schedule reminders, nicely format PDFs, and has a well-documented API for uploading line items from time tracking. I've used Stripe invoicing for this in the past, but in the situation where one has clients that use *other* payment mechanisms, IME this creates annoying reconciliation gaps. Any recommendations?

• Post #4155813

  why do you think those venture capitalists are subsidizing this thing. do you think that they are hoping that you will just increase the collective social value of the commons, and rejoice? https://mastodon.social/@juliank/116974045503028965

• Post #4124229

  RE: https://mastodon.social/@404mediaco/116986805020926545 this is one reason I ascribe more incompetence than malice to most boosters. the boosters were mad about the codeberg ban. a calculatingly evil conspiracy would understand that a brutal model collapse is inevitable for coding models if there is no fresh, nonslop option for modern code.

• Post #4104991

  in the open source world we really need some well-recognized social concept like the "attorney-client relationship" that establishes some boundaries around expectations and obligations

• Post #4100518

  WE ARE SO BACK https://www.patreon.com/creatorglyph/posts/patreon-update-164760797

• Post #4100469

  Today I am reminded yet again of one of my _favorite_ features of Fastmail https://www.fastmail.help/hc/en-us/articles/1500000278342-Server-names-and-ports#h_01H95ZZXYRR7J895AFBSH5S5KX

• Post #4077918

  really really do not want to turn my blog into an all-AI-skeptic all-the-time hate-fest but this bothered me so much that I couldn't help but write it. I am not going to edit it. if you get some value out of it let me know but otherwise I might just delete it now that it's out of my system https://gist.github.com/glyph/0c7b7d51449a1bbedae99f9422bc180f

• Post #4075747

  As a *technology*, ionizing radiation isn't exceptional. It's not exceptionally wicked. It's not exceptionally good. Take away the accompanying risk of global thermonuclear war, and we'd call radiation's applications "plug-ins" and we'd use them and abuse them in the same way we've used every other technology.

• Post #4054945

  Hello #emacs friends: on some relatively recent update, #flycheck stopped displaying error messages reliably. After some investigation, I have discovered that the culprit seems to be `(customize-set-variable flycheck-display-errors-function 'flycheck-display-errors-via-eldoc) ; after switching to `(customize-set-variable flycheck-display-errors-function 'flycheck-display-error-messages it works again as expected but (obviously) with no colors any more. Is this an issue with eldoc or flyc...

• Post #4050481

  one day we will have safety regulations written about “AI”, but, like all regulations, they will be written in blood. when I abstain from “AI” or suggest to you that you do the same, it is in large part because I would rather they not be written in my blood, or yours.

• Post #4044544

  when will the democrats finally learn that publicly arguing about which members of their coalition they are willing to feed into a wood chipper in order to not be “too far left” on “culture war issues” is a losing strategy that makes them look like useless Quislings to their allies *and* to their enemies

• Post #4044511

  Please do not add genAI images to punch up your writing. You might think that it adds a nice little bit of visual pizazz to your content-marketing piece, but what you're actually doing is *making it look like content marketing* rather than a useful resource. To the extent that content marketing is an effective tactic, it is because you build trust with the customer by providing them valuable information. A genAI turd plopped on top of your writing is a signal that it will be worthless slop.

• Post #3998810

  a summary of my response to the OpenAI/Hugging Face incident: If you run os.system(random.choice(["say --voice=hacker 'I’m in'", "rm -fr /"]) it deletes your filesystem, and then you breathlessly claim that this is a dangerous alien consciousness, that is not a demonstration of an offensive cybersecurity capability, that is not even an injection attack, that is you stabbing yourself in the thigh with a pencil and then telling everyone that pencils are too dangerous t...

• Post #3966096

  Folks, I say this with love, because it has been an issue for several media personalities whose work I really enjoy and appreciate: An audience can forgive scuffed video but you really, really need to get your audio situation sorted out if you want people to keep watching & listening.

• Post #3964989

  the things I want from "siri AI" that I'm not going to get: - "here's how I pronounce this" so I can launch things like "Authy" with a voice command (try it) - disable _all_ cloud interaction, even if it makes some things less reliable - a reference sheet of commands so I can understand its capabilities - explicit preference for "respond only to my voice" - error handling in Shortcuts none of these things involve an LLM, and attempting to cram LL...

• Post #3930865

  when you think about it an RST packet is a kind of existential horror

• Post #3902523

  give a man a fish, feed him for a day. teach a man to read the subreddit about aquarium maintenance, bankrupt him for life

• Post #3898867

  It is no longer safe to turn off your computer.

• Post #3870249

  almost every group that we conceptualize as a big evil organization that is ruining our lives is in actuality a patchwork of smaller organizations, at war with itself. there are people trying to do the right thing, and the wrong thing. this is particularly true of corporations, since *most* people are employed by corporations.

• Post #3869249

  starting to think about sandboxing my python apps, as everyone should probably do, but the only thing I've dug up so far that's a "real world example" is https://github.com/davidfstr/Python-in-Mac-App-Store which is 13 years old and set up for Python 2.7, so, not _super_ relevant. has anyone submitted a Mac App Store python app in the last 5 years or so?

• Post #3841374

  RE: https://social.treehouse.systems/@mgorny/116922216150959614 Trying really hard not to come to this conclusion and to focus on the fact that we must cultivate our garden