Remote

Glyph

@glyph@mastodon.social

mastodon 4.8.0-nightly.2026-10-06

Gender: he/him
Writer: https://glyph.im
Coder: https://github.com/glyph
Founder: https://twisted.org
Maintainer: https://pypi.org/user/glyph
Streamer: https://twitch.tv/glyph_official
Speaker: https://pyvideo.org/speaker/glyph.html
Busker: https://patreon.com/creatorglyph https://github.com/sponsors/glyph https://ko-fi.com/creatorglyph
Campaigner: https://superswingdistricts.org
Public Figure: https://en.wikipedia.org/wiki/Glyph_Lefkowitz
Tooter: @glyph@mastodon.social@justmytoots.com

7527 Followers

359 Following

50 Posts

Joined February 06, 2017

posts:

https://blog.glyph.im/

disclosures:

https://blog.glyph.im/pages/disclosures.html

code:

https://github.com/glyph

Glyph @glyph@mastodon.social

· 1w ago

Boosted by @trending@homestead.social

I logged in to LinkedIn to see a page full of people praising DHH, got mad, and wrote this. Less than 12 hours later I got an email telling me "Your post doesn’t comply with our Professional Community Policies on bullying and harassment. It’s been removed from LinkedIn and only you can access it." Big tech is more and more mask-off every day.

608

0

451

0

Glyph @glyph@mastodon.social

· 2d ago

Boosted by @fedicat@pc.cafe

done correctly, open source is a gratitude practice which may occasionally produce software

32

0

13

0

Glyph @glyph@mastodon.social

· 2d ago

Boosted by @trending@homestead.social

RE: https://mastodon.social/@bagder/117386899276807175 boosters one month ago: the vulnerabilities are REAL. they're REAL and they're SCARY. it can INDEPENDENTLY hack stuff. you CANNOT DISMISS that it can ONE-SHOT vulnerabilities PERFECTLY EVERY TIME. prepare for DOMINATION by the CYBER-OVERMIND boosters today:

Quoting

@bagder@mastodon.social

"Google freezes open-source bug bounty program amid flood of invalid AI slop submissions" https://www.tomshardware.com/tech-industry/artificial-intelligence/google-suspends-part-of-the-oss-vrp-bug-bounty-program-due-to-an-influx-of-invalid-ai-submissions-product-vulnerability-submissions-ended-october-1

Open quoted post

mastodon.social

daniel:// stenberg://: ""Google freezes open-source bug bounty program am…" - Mastodon

24

0

25

0

Glyph @glyph@mastodon.social

· 4d ago

Boosted by @trending@homestead.social

Me: I will use "radium water" as an analogy for how horrible AI's potential subtle long-term effects are, because the general public in 2026 can easily understand this analogy. The general public in 2026: https://futurism.com/future-society/alternative-health-influencer-drinking-radioactive-jug-revigator

Alternative Health Influencer Boasts That He's Been Drinking From a Radioactive Jug for 250 Days Straight

Alternative Health Influencer Boasts That He's Been Drinking From a Radioactive Jug for 250 Days Straight

Holistic influencer Matt Roeske recently bragged that he's been drinking from an irradiated jug for the better part of a year.

24

1

20

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

I believe lots of things about AI but we could all avoid so much pointless bickering if some OpenAI executive simply went to prison every time ChatGPT abetted a homicide. You only have to do this 2-3 times before people start taking shit really seriously!

116

4

51

0

Glyph @glyph@mastodon.social

· 1w ago

I realize I'm in a strong minority here but I just _cannot_ perceive submitting LLM-generated work to a volunteer project as anything but an extremely rude and thoughtless thing to do. I _know_ that the consensus social etiquette is that I'm supposed to have a prominently-posted policy statement that this now-generally-accepted-as-normal thing is actually considered bad in this specific place, but it remains jarring for me every single time.

186

7

51

1

Glyph @glyph@mastodon.social

· 1w ago

Replying to

It's not even "I wrote some software and put it in a product and the product did a harm and I'm maybe fractionally culpable in a way that's really hard to apportion" any more. It's just "I set up a computer to do crimes and I spent ten million dollars on electricity to prove it could do crimes and it did a crime. Who could have known!"

113

3

41

0

Glyph @glyph@mastodon.social

· 1w ago

Boosted by @joe@f.duriansoftware.com

New Blog: What Would A Serious AI Product Look Like? Every “AI” tool is missing critical features that you would need if you wanted to do real work with them. What would those look like, if they existed? https://blog.glyph.im/2026/09/serious-ai-product.html

What Would A Serious AI Product Look Like?

Every “AI” tool is missing critical features that you would need if you wanted to do real work with them. What would those look like, if they existed?

101

17

88

6

Glyph @glyph@mastodon.social

· 6d ago

Boosted by @trending@homestead.social

Taylor Lorenz is preparing for her "why I left the left" right-wing grifter pivot. Do not trust her. https://www.thefp.com/p/two-drinks-with-taylor-lorenz-savodnik (If you do not know "the free press", it is Bari Weiss's right-wing false-balance propaganda outlet, so talking to them *at all* is already suspicious. Talking to them about how you "wouldn’t say [you're] particularly woke" is a dog whistle so loud my ears are bleeding.)

Two Drinks with . . . the All-New Taylor Lorenz

Two Drinks with . . . the All-New Taylor Lorenz

Taylor Lorenz, social media’s best-known gadfly, insists: ‘I haven’t changed what I believe. I’ve just gotten disillusioned about the political landscape. There’s not a clear political group I map onto now.’

45

1

25

0

Glyph @glyph@mastodon.social

· 1w ago

Boosted by @jwz@mastodon.social

Replying to

There's a whole legal/philosophical argument that software is a form of speech, and we don't punish harms done with speech, so we can't punish harms done with software. But that's not true at all either! There are very strict limits for speech, even in the USA where the limits are as loose as they can get! Fox News had to pay $787.5 million dollars! Fraud is still a crime even if you do the crime entirely with words!

64

1

8

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

The HuggingFace hack is like the most overt, offensive example of this pattern. "We did a bunch of federal crimes. Oops I mean A ROBOT did a bunch of federal crimes, I'm touching my nose, you can't put me in jail for a premeditated CFAAA violation"

60

3

12

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

These are not crimes of passion by the downtrodden, it's calculated exploitation where everyone involved has more than enough time to carefully consider the ramifications of their actions. This is the circumstance in which deterrence really works! Everybody at a big corporation is trying to color inside the lines at all times, but they have learned that "the lines" our society has drawn will allow them to do nearly infinite crimes as long as they are performed at one level of remove

58

3

15

0

Glyph @glyph@mastodon.social

· 5d ago

Boosted by @jwz@mastodon.social

Replying to

cc @mttaggart@infosec.exchange

18

0

8

0

Glyph @glyph@mastodon.social

· 1w ago

New Blog: Who is open source about? https://blog.glyph.im/2026/09/who-is-open-source-about.html

Who Is Open Source About?

Open source comprises a complex social web of ongoing relationships, not a simple one-way “gift” of value from maintainer to user.

66

0

68

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

I'd like to ask CPython, today, to stop incorporating slop. This discussion would likely go nowhere. I cannot imagine a way to have it productively. But there's a future (this *specific* one isn't likely) where, for example, NYT wins a devastating blow against Anthropic, something comes out in discovery about proprietary code as well as pirated books in the training set, somebody sues the PSF and then someone needs to review and potentially reauthor every commit from Python 3.14 to 3.19.

43

5

10

0

Glyph @glyph@mastodon.social

· 5d ago

Replying to

of course they're fucking this up, why wouldn't they fuck it up, they're just inventing new things to fuck up constantly

9

1

3

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

I do still (perhaps irrationally) believe that we can potentially redeem the community and rehabilitate everyone currently acquiescing to the slop bots but the thing that I'm struggling to adjust to is that I don't think I will be alive to see it, and some of the setbacks along the way are going to be just excruciatingly devastating.

32

6

3

0

Glyph @glyph@mastodon.social

· 1w ago

Boosted by @jwz@mastodon.social

Replying to

@rmi@cloudisland.nz it still seems like a weird thing to say! again: lots of people hopping on the “web” bandwagon in the early 2000s, yet nobody felt the need to go on the goddamn nightly news and proclaim “you have to admit, netscape is real”

29

3

9

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

Imagine the impact this would have on the community. Imagine trying to raise money required to re-litigate 5 years of work from a hundred contributors. Would Python continue to exist? Would we just have to immediately vaporize all progress made in the interim, start over at 3.14 and desperately fix 100 CVEs at once, carefully _not_ backporting all the fixes? Would every ecosystem package author need to join in to this all-hands-on-deck effort?

26

4

3

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

like… LLMs… do… exist. they do stuff. interesting and confounding stuff and stuff worthy of research, if not, in my opinion, net-utility useful stuff. machine learning and previous iterations of "AI" are unambiguously useful in various contexts. I never hear skeptics saying that transformers or neural networks or whatever "aren't real". and yet the industry doth protest WAY too much.

28

4

7

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

Now imagine this happens to Linux. It's impossible to redeploy half-a-decade-old kernels without automated help for the security fixes. So, to "help out", Microsoft releases a version of Windows for the Cloud with an up-to-date WSL on it, so you can keep using your Linux software while you migrate to the "legally less risky" windows API. Windows is still equally full of slop of course but nobody can prove it without public source and they have a legal team, so you could actually install it.

22

3

3

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to on mastodon.social

chat is it good for your stock valuation if every user of your product insists, intensely, that they're not hallucinating and that your company actually exists

24

10

3

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

me shouting down every other panelist on Squawk Box: "You don't understand. My blog is REAL! It's real. you have to ADMIT it's real. You can sign up for my patreon right now, and give me actual dollars. It's REAL guys. It's just not SERIOUS to suggest that my blog IS NOT REAL"

17

9

2

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

sam altman rocking back and forth, muttering to himself "chatgpt is real, and he is strong, and he is my friend"

16

6

3

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global @zzzeek@fosstodon.org @hynek@mastodon.social I've also explained, both in that article, in general posts, and to you personally, why people *massively over-perceive the benefits and under-perceive the costs*. it's too early to have any data on this but neurodivergence is theoretically a huge risk factor for AI psychosis

psychologytoday.com

13

8

4

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

I swear I issued this query because I was getting annoyed at hearing boosters say this so often and was trying to find a precedent for someone using this phrase to describe literally anything else. on page 9 of the google results I found a tiktok about breeder reactors.

14

8

1

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global I have thought a lot about this in the past decade. there's a lot of validity to what you're saying, but I think there's a really important additional fact, which is perhaps the root cause of this perverse dynamic: attempting to evaluate behavior on a case-by-case basis, particularly when evidentiary rules exclude "out of context" stuff, grants an asymmetric advantage to bad actors because they can pantomime good behavior in context while having obviously nefarious goals

15

8

2

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global circa … maybe around 2010, I had a very strict "always evaluate only the behavior in evidence, in community spaces" approach to moderation decisions and I absolutely got my soul crushed by fascists repeatedly, intentionally attempting to grind down shared spaces by constantly line-stepping and rules-lawyering nonsense. gamergate, CoC debates, you name it, just constant bad-faith bullshit that was technically "within the rules"

14

7

1

0

Glyph @glyph@mastodon.social

· 2w ago

RE: https://wandering.shop/@phildini/117310939687993407 Folks let's get some more boosts on this and connect to a Grafana expert, we can't have this AI harm mitigation strategy absolutely face-plant on the _first try_. (I am Very Bad At Grafana™ so this one's beyond my ability to just jump into the breach on, which is probably a good thing)

Quoting

@phildini@wandering.shop

Hey, does anyone here consider themselves Good™ at #Grafana and wants to spend like half an hour on a call with me helping me be Good™ for the sake of @civicband@sfba.social ? I'm trying the @glyph@mastodon.social suggestion of "bug your friends instead of asking AI"

Open quoted post

phildini: "Hey, does anyone here consider themselves Good™ a…" - The Wandering Shop

20

1

25

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

now I'm trying to find where critics say "the technology isn't real" but it's all still just boosters doing fake nuance, just negated. "not because the technology isn't real"

11

5

1

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global @zzzeek@fosstodon.org @hynek@mastodon.social also, it is extremely rich to say this specifically to me, the person who has been flogging my extremely in-depth article about tradeoffs for an entire year now https://blog.glyph.im/2025/08/futzing-fraction.html

The Futzing Fraction

At least some of your time with genAI will be spent just kind of… futzing with it.

10

9

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@zzzeek@fosstodon.org @eevee@slime.global @hynek@mastodon.social look man, I can't stop you from using it. I don't want you to prove to me that "it works". You are a gambling addict, scoffing at skeptics because you insist that what they're saying is that jackpots never happen. It's not what I'm saying and you're not going to convince me or any other sketpic by showing me your successes, if you wanted to even begin to convince me the starting point is a record of where and how it failed, where it wasted your time

9

7

2

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@zzzeek@fosstodon.org @eevee@slime.global @hynek@mastodon.social Instead what you are doing is addiction shame-spiral behavior. "But I analyzed some logs and doing coke beforehand saved me a TON of time, what do you have to say about THAT!" is a fact-free assertion that I couldn't possibly address logically, it's just meant to browbeat me into admitting that what you are doing is OK, that it's fine to slowly start depending on cloud-hosted full-fat Claude for every aspect of your life. It isn't! I'm not!

8

2

1

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@zzzeek@fosstodon.org @eevee@slime.global @hynek@mastodon.social I would be much more interested, personally, in hearing about the personal psychological safeguards you're putting in place or the dependency mitigations you have, ways you're engaging in deliberate practice to avoid the inevitable collapse when the financial crisis hits, then about how great it is. Brandolini's law applies here, debunking any particular chatbot application is 100x harder than boosting it.

8

6

1

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

on the very last page (16) there's an article about TSMC's chip technology, but that doesn't count, because it's technology for making AI chips

8

7

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@zzzeek@fosstodon.org @eevee@slime.global @hynek@mastodon.social I appreciate that my opinion means so much to you that you feel compelled to keep bringing this up so I won't think poorly of your decisions, but your arguments are anti-convincing me.

7

5

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@zzzeek@fosstodon.org @eevee@slime.global @hynek@mastodon.social I don't know that you'd convince me with that but it would at least be consistent with the position that there are big enough gains to be worth some hassle and some risk, that you take the risks seriously.

5

3

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global @zzzeek@fosstodon.org @hynek@mastodon.social and to mike's original point, this is, itself, harm that every single user experiences. either you are being a sucker, embarrassing yourself publicly like this as the chatbot tricks you, or you are checking its output and it's "actually useful" which just means you are being subjected to the constant risk that a momentary lapse in vigilance will lead you to more reliance, that the reliance will increase in a loop you can't perceive. that is also not good for you

5

9

1

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global yeah, and even in less extreme cases, this type of approach presents some pretty terrible incentives. wreckers will start trying to label everyone they even mildly dislike. Read Some Theory jerks will label every moderate who doesn't have time to develop their own unique synthesis of Kropotkin and Marx as an enthusiastic genocidal fascist. arguments which SHOULD be about behavior instead center on ideology

6

4

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@zzzeek@fosstodon.org @eevee@slime.global @hynek@mastodon.social If you really just thought the technology was good and was a benefit to open source, you might be saying something like: I hear you on the concerns, but you could set up a local Pi instance with Qwen. Here's how you set up a VM to isolate it so you can check its work and not give it network access. Here's how you measure the benefit vs. the time it's wasting.

4

4

1

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@hruske@mastodon.green I earnestly don't know what's going on. I don't know who these people — and there are a LOT of them— think they're responding to. I am doing some gallows humor here but I am seriously wondering if something about feeling defensive and interacting with a chatbot leads one into some common thought pattern that produces this exact phrase, and I'm witnessing an artifact of some mass psychosis

3

2

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@jwz@mastodon.social vile. I want to quip about this but I got nothing; it’s just vile. Hopefully every single one of these cowardly dimwits is unseated soon.

1

0

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global the problem with trying to penalize "constantly testing the rules" is that it is _very_ easy to muddy the waters and do a bunch of stuff that massively stresses everyone out and is kind of awful but is hard to overtly prove is a transgression. something like the Olivia Hill rule is probably the best way to break this logjam but socializing that requires a surprising amount of effort

1

3

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@eevee@slime.global (cue a lengthy complaint about how that's just, like, your opinion, and doesn't everyone have the RIGHT to be in the space, it's not like I BROKE any RULES, etc etc etc. and yeah the immune system of the space needs to have a way to recognize this)

1

1

0

0

Glyph @glyph@mastodon.social

· 1w ago

Replying to

@prism@infosec.exchange I don't know. It seems like it would be hard, but they have some theory of how they are going to do indemnification of all their customers that presumably some lawyers actually looked at, and they've had it since the beginning: https://blogs.microsoft.com/on-the-issues/2023/09/07/copilot-copyright-commitment-ai-legal-concerns/

blogs.microsoft.com

0

1

0

0

Glyph @glyph@mastodon.social

· 6d ago

Replying to

@jwz@mastodon.social I wish I had something specific to offer, but the best I've got is that there were a bunch of code-signing changes specific to screen savers in macOS 26; does this have an ad-hoc signature or a proper public developer ID one?

0

1

0

0

Glyph @glyph@mastodon.social

· 5d ago

Replying to

@nixCraft@mastodon.social this is a good take but let's remember that Yann LeCun was "Chief AI Scientist" at Facebook from 2013-2025, so this is a duck that comes pre-milkshaked

0

0

0

0