Thomas A. Fine
banner
thomasafine.bsky.social
Thomas A. Fine
@thomasafine.bsky.social
Unix guru, Internet hacker, security curmudgeon, software engineer, politics fanatic, media critic, bike nerd, astronomy fan, history buff.
So for example, if I asked it about the 2025 NYC EO #52, it would give me a description for an entirely different EO. Usually randomly picking some other EO that got a lot of attention.
January 2, 2026 at 11:18 PM
And just yesterday, when I was researching all the undone executive orders from Mayor Adams, over and over again Google's AI summary would give me a description of a given executive order that was 100% wrong. It was unable to differentiate EOs by number.
January 2, 2026 at 11:18 PM
People are better off thinking of LLM answers as random. The number of people I see who "fact check" by asking an LLM is frightening.

Whenever there's a controversial thing, you can literally find different people getting opposite fact-check answers from the same LLM.

Functionally random.
January 2, 2026 at 11:13 PM
Every single story about large language models should start with things like:
"Grok engineers..."
"OpenAI developers..."
"Gemini marketing..."
etc.
January 2, 2026 at 11:06 PM
OK, "under no obligation" is the disingenuous framing here. When is anyone ever under an obligation to spend money in any specific place? It's a nonsensical thing to say.

Meanwhile boycotts, good or bad, are about building an obligation to harm some enterprise.
January 2, 2026 at 10:55 PM
Disingenuous framing? It's a real actual example of targeting Jews in America who had no connection to Israeli political or military machinery.

And I didn't say boycotting was always bad. My point is just that it is not automatically good. And can be bad if done thoughtlessly.
January 2, 2026 at 10:51 PM
Responding to your answers is not spin. It's just responding.
January 2, 2026 at 6:08 PM
I'm making good-faith efforts about how bigotry functions (and not losing anything).

You're getting defensive and lashing out, not making good-faith arguments.

Here's the thing: if "I am not a bigot" is part of your core identity, you never check yourself. And hence, you risk becoming a bigot.
January 2, 2026 at 6:03 PM
Oh boy.

The holocaust happened not because most Germans "hated Jews". But because most Germans avoided the uncomfortable politics that had been spun around Jews... by avoiding Jews.

"I avoid them because I'm just not political."

Welcome to the machine.
January 2, 2026 at 5:52 PM
Free speech and anti-semitsm (or any bigotry) are not mutually exclusive. if you can't understand that, you can't understand anything.

If you avoid a restaurant because it's Jewish-owned, that's 100% as bigoted as avoiding a restaurant because it is owned by any other minority.
January 2, 2026 at 5:48 PM
Targeting domestic Jews to address a perceived wrong happening overseas is not antisemitism. Right. Keep telling yourself that.
January 2, 2026 at 5:36 PM
As a kosher restaurant, anyone could of course eat there, but for Jews who keep kosher, it was one of a very small number of places to go out to eat.

So the impact was absolutely disproportionately against local Jews, far more than anyone else.
January 2, 2026 at 5:30 PM
If you need a gut-check about whether something is anti-Semitic, use the substitution method: find a parallel behavior and see if it is ok.

Would it be ok to put a Russian immigrant that runs a restaurant in the U.S. out of business, because you're "boycotting" Russia for the war in Ukraine?
January 2, 2026 at 2:56 PM
No. People keep chanting this mantra like it protects any and all behaviors. When "boycotts" solely impact Jews in the U.S., while having essentially zero impact on Israeli policies, then this is bigotry, not boycotting. Here's an example of what I'm talking about:
bsky.app/profile/thom...
January 2, 2026 at 2:46 PM
Here's the public reaction to my hack. (Note that they were wrong, root access was not required.)
January 2, 2026 at 6:57 AM
And not long after my hack, a colleague did in fact implement one of the earliest ever firewalls. Although Doug Karl's firewall seldom gets credit for being the first, I think it may well have been the first. (Doug went on to invent WiFi, and then get largely forgotten for that also.)
January 2, 2026 at 6:38 AM
You could also make the case that my biggest impact on the Internet was in 1988 when I hacked the rexd protocol and broke into some places. Within a small niche of the Internet, this hack was a turning point where people began to realize that Internet security might be an important thing.
January 2, 2026 at 6:38 AM
And not all armchair physicists in the bike community accept these ideas, but the industry itself embraced them.
January 2, 2026 at 6:27 AM
And those two articles were important, because they challenged previous understandings about energy loss in bike frames.

hea-www.harvard.edu/~fine/opinio...
hea-www.harvard.edu/~fine/opinio...
A Stiff Frame is Good to Find
hea-www.harvard.edu
January 2, 2026 at 6:27 AM
I did also single-handedly redirect bicycle frame building with a pair of articles exploring why both stiffness and compliance were important for bike frame design.

I didn't coin the phrase "laterally stiff vertically compliant", but I was the first to describe the underlying ideas.
January 2, 2026 at 6:27 AM
Bind: Address Already in Use
hea-www.harvard.edu
January 2, 2026 at 6:27 AM
But probably more important, and longer lasting, was a technical article for developers on the bind system call, and the "address already in use" error. I provided the definitive explanation still cited today.
January 2, 2026 at 6:22 AM
For a time, I was the biggest source of information on the web for ISDN back when ISDN mattered.
January 2, 2026 at 6:22 AM
But setting failures aside, the Usenet FAQs, a now forgotten archive of text documentation auto-converted into HTML that I ran was huge for a while, won an award, and was influential on future services.
January 2, 2026 at 6:22 AM