ulyssessword
No bio...
User ID: 308
ulyssessword
2d ago
I would also be curious what gets journalist to write things like these articles. You can read the article and by her own language realize she is writing in a specific style that she knows she doesn’t have the goods. But still rights the article anyway. Is it just needing a paycheck?
My take is that they get the idea for an article, research it, then write out their idea (regardless of their findings). That's very different from finding an interesting topic, researching it, then writing out their findings (regardless of their initial idea).
For example, Machine Bias could've come about like:
- There's a new recidivism predictor. Tech is bad and courts are racist, therefore this will make racist decisions in the courts.
- It shows some predictive power, and is well-calibrated (recidivism rate is approximately (20 + 5 * risk score)%, with almost-entirely-overlapping error bars between races). Maybe there's something about what it's replacing as well.
- Here are some outliers, some stories about racism, and a few charts and tables showing a complex measure nobody has ever cared about before or since. Voila, point 1 stands.
ulyssessword
2d ago
revealed in Propublica
ProPublica has zero credibility, and I will bang this drum every time someone cites them favorably. They damaged my faith in journalism more than anything else has or will. Also a perfect example of what we discussed last week: They are members in good standing of the Journalism Club, which tells me what their standards are and how they deal with deceptive and manipulative content. It's even "notable reporting".
1, 2, 3 are my posts I could find easily, but there are more out there.
With that out of the way, let's look at the article.
Leaders of the prison gang known as Los Tiburones, or the Sharks, were selling drugs to inmates not only for money, but for their votes...Investigators had gathered solid evidence of election fraud implicating both inmates and staff,
Alright, cool.
But as federal prosecutors prepared an indictment against the inmates and staff in November 2024 — just days after Trump won the election and González-Colón clinched the governorship — they received a surprising directive. Their bosses in the U.S. Attorney’s Office for the District of Puerto Rico instructed them to exclude the voting-related counts against the inmates and all charges against the prison staff, an investigation by ProPublica found.
"Just days after Trump won the election" is well before any real actions were taken to transition power from the Biden administration to Trump. Why did W. Stephen Muldrow (appointed under Trump I, dropped and immediately reappointed under Biden, and maintained in Trump II) do that? It's possible that Biden appointed someone disloyal, but it's also possible that it's completely mundane.
charging 34 inmates and associates
That's it?? I know that the election fraud offenders don't have to be a subset of the drug offenders, but it certainly suggests that it's a smalltime operation.
In court documents tied to a different case, in October 2025, a magistrate judge mentioned “an unrelated white-collar investigation involving the Governor of Puerto Rico.” Muldrow’s office responded in a filing, stating, “There is no white-collar investigation (or any other investigation) of Puerto Rico Governor Jenniffer González-Colón.”
"Involving" and "Of" are two different words.
González-Colón has not been charged with a crime.
Wow. Such news.
Raquel Rutledge (the author) has not been charged in the disappearance of Jack and Lilly Sullivan. This is 100% factual and you can check the public records if you doubt me.
the Office of the Director of National Intelligence seized the voting machines
How is that even tangentially connected to this scheme? All a voting machine can do is properly and accurately (or improperly and inaccurately) record what is entered into it. The machines don't have mind-reading equipment that can distinguish a coerced vote from a free one.
Over time, federal prosecutors say, several of these [nonprofit advocacy] groups operating in the prisons evolved into violent criminal organizations such as Los Tiburones and Ñetas,
I couldn't find any evidence that Los Tiburones "evolved", as it appears to have always been a criminal group. The Netas started as a legit advocacy group, and still use it as propaganda.
the party won 83% of the inmate vote, according to a ProPublica tally of voter returns on the State Elections Commission’s website.
Inmate votes were especially key in the 2024 gubernatorial primary as González-Colón, a longtime New Progressive Party member, was challenging the incumbent governor of the same party.
She won the primary by fewer than 30,000 votes, according to the State Elections Commission. Local news reports said that an estimated 5,000 prisoners voted territorywide.
No evidence of what the within-party split was in the primary: Extrapolating to 6000 prisoners total, 5000 support the Progressives, and of those 5000 an unknown number supported Colon with the remainder supporting her opponent(s) within the party primary.
(Fake edit: Later in the article has "...being pressured to vote in the primary — some for González-Colón and others for her opponent, Pedro Pierluisi.". Why wasn't that in the Primary section of the article? Oh wait, they moved from "Colon is benefiting" to "Prisoners are compelled" and expected that you couldn't cast this point back in time to where it would undermine their argument.)
Political analysts said rumors have swirled over the decades about coercive tactics being used to mobilize the prison vote, raising significant questions about the extent to which that support comes in exchange for favors from the ruling party.
This time was different, sources said. They had evidence.
...of something else, not what was mentioned in the previous paragraph. They're just "deep into investigating a potential..." for that part of it.
A relative of one of the prisoners told ProPublica that inmates had to show their ballots to gang leaders when they voted to avoid punishment.
God fucking damn it. There's a second breach of election security happening? Fix that, and vote buying becomes a pure game of trust. Given how trustworthy I find the prison population, I'd guess it would immediately kneecap any election influence operation.
ulyssessword
3d ago
Do you have an argument, or just a bare assertion?
ulyssessword
4d ago
A wind blows and a bridge falls, we can trace back what happened and why. An LLM gets stressed, allegedly; what do we trace back? Why does it happen?
Model weight activations, etc.
From a practical engineering perspective, model emotions are easily observed and kind-of understood. The question is whether that concept is a valid match for human emotions. For the bridge example, we could say that it "feels like dancing" in the wind and build practical mitigations, but it would fail the second step because the bridge's so-called feelings don't match with a person's.
ulyssessword
4d ago
we can look at levels of cortisol and the rest of it to measure stress.
What makes us think cortisol etc. is associated with stress?
It's something like the researchers finding (or creating) stressful situations, and then measuring increased cortisol in the brain. Then following it up by enhancing/suppressing cortisol and seeing if it causes/prevents stress-like responses. It's the same with LLM probes: researchers classify some situations by apparent emotion, see the model activation similarities, and name those clusters of activations in a simple way after testing them.
Why should an AI get stressed about "impossible task with high stakes" given that we are betting the farm on AI being smarter and more capable than mere humans and able to handle the problems that get our knickers in a knot?
I think we slipped down a level of discussion here.
It's like asking whether moderate wind should cause a bridge to collapse: It's a design flaw, but it follows from the laws of the universe. If we're betting that much on a bad design, then that sucks because the bridge will fall and the LLM will be stressed.
That question has almost no bearing on the baseline question of whether LLMs get stressed (or have any other emotion).
ulyssessword
5d ago
And there is no reason to think that Claude feels anything at all.
There may not be sufficient reason to think Claude feels emotions, but there's unquestionably some reason to think that.
If you put the chatbot in a situation that would create anxiety in a human (e.g. an impossible task with high stakes), then it will write statements that are consistent with anxiety and have degraded performance in a similar way that an anxious person would. "It's anxious" isn't the only explanation for that behaviour, but it's a possibility worth considering.
You can also peek (one level) under the hood and look at the activations of Claude's emotion concepts. It feels activates its model weights in a pattern consistent with certain emotions in situations that you would naively expect it to feel those emotions, and artificially activating (or suppressing) those activations causes changes in behaviour consistent with increases (or decreases) in the emotion.
Interviews, job performance, psuedo-brain scans, and pseudo-drug responses all point to Claude having emotions. It's not a bulletproof argument, but it's one worth taking seriously.
ulyssessword
7d ago
The common version of that is called Gell-Mann Amnesia. Unfortunately, there aren't great ways to avoid it.
ulyssessword
7d ago
At minimum, I'd think, she'd use, fewer, commas,,,
ulyssessword
8d ago
Try to think of a way of hiring or promoting people that does not benefit one group at the expense of another.
Simple: give up on meritocracy and equal protection, and set a quota by race/sex/etc. If you can't tune it perfectly (e.g. 5/10 board members can be women, but other areas don't have strict categories or nice numbers), then be sure to err against non-protected groups.
It's not good, but it is almost as simple as figuring out what the criteria is and then following it.
ulyssessword
8d ago
It's outdated by now, but exit polling of the 2016 election had Black voters favoring Clinton more than Republicans favored Trump. It was literally more true than "I don't want to be assumed to vote Republican just because I'm a Republican" in that case.
ulyssessword
9d ago
At first I didn't like the constant cuts which I took as downstream from ai generation but it does seem consistent with the action genre so maybe it was a choice and not a technical issue.
A bit of each, probably. Current-gen AIs can match the Action genre, which is why you're seeing action movies before anything else. In an alternate world where AIs could do characterization, dialog, etc. before visual fidelity, boobs, and explosions, you'd see different movies coming out first.
ulyssessword
9d ago
96% here, and my exception
ulyssessword
10d ago
How much can I make in a year of substandard (or sub-excellent) healthcare? How hard is it to think up a reason to get more care and therefore more cash?
Currently, the patients bear some cost for seeking healthcare: even if their out-of-pocket expenses are zero, they still have to invest time, effort, and discomfort. Under your idea, seeking care becomes profitable. Probably more profitable than working at a job for some people.
ulyssessword
11d ago
You can't kick someone out for something like murder before they murder obviously, this isn't Minority Report.
Yes you can! The Zizians were kicked out for pre-crime. No superhuman precogs required.
What bullet am I biting...
Sorry, wrong subject for that sentence. I have to bite the bullet and accept collective responsibility for my countryman's actions if I want to be consistent. You have already set yourself apart.
I know that there's an incentive to give myself a pass by claiming he isn't part of my group, but that would be a lie. He's a scotsman, and scotsmen murder.
Then it's not collective responsibility if you allow all the people who aren't responsible and don't actively support bad things to dodge blame!
That's tautological. The people who aren't responsible aren't responsible while the people who are, are. We simply set different boundaries. People can dodge blame by genuinely not being blameworthy.
Yes exactly, and all what I've given is motivated reasoning I've seen people engaged in!
They're wrong. There's not much more to add to that.
it's generally related to their own personal experiences feeling uncomfortable and not the actual thing in question "I'm responsible for what someone else did".
I've seen variations of "We failed at..." and "I wish I could have done something about..." alongside other implicit claims of responsibility. "I'm responsible for..." isn't a common saying in any part of society (unless it's something boring like a task at work), so I'm not surprised it isn't explicit here either.
If you're the head of the "Dog Lovers club" and kick someone out, they can just go make their own "Dog Likers" club.
Consider Autism Speaks. They are roundly and consistently criticized by other autism advocacy organizations for their stances. If they do something, it doesn't reflect on those other organizations because they have been kicked out of the cool kids club.
ulyssessword
12d ago
Not just yours, it's a common misrepresentation.
ulyssessword
12d ago
He's very publicly in favor of an international alliance of USA, EU, Russia, and China firebombing rogue data centers.
Airstriking. He chose his language deliberately so it doesn't get confused with a call for individual action.
ulyssessword
12d ago
"At what p(doom) do you saw your own leg off?"
Why are you so certain that firebombing is the most effective strategy available? It doesn't have a great history, to say the least.
ulyssessword
12d ago
If a person believes...
You're missing a step: They have to claim that they're a bad group, then actually kick them out so that they aren't a part of the larger group anymore. Cut them off from funding, stop inviting them to events, denounce them, etc.
These two ideas go in hand and hand. A belief in collective responsibility leads to denial of someone in your own collective ever doing any wrong. A Scotsman who believes in collective responsibility will have to say "That murderer isn't a real Scotsman!" because admitting he is a Scotsman = Scotsmen are murderers in that logic.
Whereas if I was a Scotsman, I could say "that murderer is a scotsman too, but he's not me, so I'm not responsible".
You have to bite the bullet and accept it, otherwise you're a hypocrite. I'd say "It's a tragedy that brings shame on us all". Now, if the murderer was a recently-deported immigrant, I could say that the murderer isn't a scotsman because he isn't.
So what? If collective responsibility is resolved by saying "I don't support that" or not doing the thing in question, then collective responsibility doesn't make sense to begin with.
It's resolved by actually not supporting it, and by opposing it. Simply saying something might not be enough.
Sure you can kick them out of a specific rationalist group, but you can't kick them out of "rationalist".
Yeah, "The left as a whole" has a lot less control than "Democrats", and that causes less responsibility to fall on the broad movement than the specific organization. In the absence of new evidence, I'd draw less of a connection to the Left for this assassin than I do to the Democrats for Sam Brinton.
As I've wrote elsewhere, this is incredibly easy and simple to dismiss by anyone motivated to put down the anti AI movement that this is just plausible deniability of stochastic terrorism.
Yeah, motivated reasoning can get you pretty much any result you'd like. If someone's dedicated to being wrong, I certainly can't stop them.
I've never really seen a person go "I believe in collective responsibility, someone in an adjacent group of mine did something bad, I am responsible".
I've never seen it cash out like that, but the general theme isn't uncommon in my experience. It's usually more like "I won't join that group because I would be supporting their bad actions" or "I regret joining that group...". People don't share their personal failings very openly.
I'll give one personal story, with all the serial numbers filed off. I was playing sports with an aggressive teammate, and he injured an opponent. I felt bad, the coach felt bad, my teammates felt bad. We accepted collective responsibility, the coach pulled him from the rest of the match, and all of us increased our focus on sportsmanship and fair play in the next few practices and the rest of the season.
We could've partially dealt with our responsibilities by kicking him out of the team, but it turned out to be unnecessary.
ulyssessword
12d ago
Are you an anti-Trump believer, or a member of an anti-Trump activist group?
If you're actively pushing anti-Trump rhetoric, then be careful of what you're doing and what consequences it has. If you're supporting others pushing rhetoric (even if it's just social/moral support, not necessarily concrete/financial support), then you still have to pay attention.
If you simply believe it in your heart of hearts, then pretty much think whatever you want. If you don't have any power, then you can't be expected to do anything about it. You might still get hit with guilt-by-association, but not enough to matter.
ulyssessword
12d ago
Well, I guess I disagree with the government. Isn't the first time, won't be the last.
ulyssessword
12d ago
Kudos to the club, they're doing their civic duty. Maybe check on the lessons they're giving and see why extremists are attracted to them, but they covered their ass pretty effectively.
ulyssessword
12d ago
Take half a second and think about this with your brain. "My movement is only good people and bad ones are unrelated" is the conclusion someone would have if they also believe "small number of people did something bad so everyone tangentially related is responsible or guilty"". You're not arguing against me, you're agreeing with me in pointing out the flaws of this logic.
What? Maybe if they're a hypocrite.
Simple example: A large group contains a small number of baddies. Does it affect their reputation? I say: Yes, they are one group and share a reputation. You say: No, the large group's reputation is not impacted by the bad individuals.
How are those identical?
Actually my stance gives full consequences for bad things to the people who do bad things, instead of trying to absolve them.
Guilt doesn't diffuse to nothing, it multiplies. A hitman can't get off from a murder charge because he was just doing a job. Instead both the hitman and the purchaser are guilty.
Similarly, the rioters throwing molotovs, the protesters giving them cover, and the pundits encouraging them all carry blame for the damage in the BLM riots.
Like how the "rationalists" had the niche cult of "zizians" who murdered people. But would you blame someone like Scott Alexander or Yudkowsky for those murders?
They split from the group, CFAR kicked them out, they had the cops called on them, and had (at least one) callout post against them. This is the appropriate reaction. They weren't members in good standing, so the blame is very heavily dampened.
That Yud can claim all he wants that he doesn't want houses firebombed, but his anti AI rhetoric lead to this so he's guilty too.
He does claim that he doesn't want houses (or data centers) firebombed. Repeatedly and consistently, both before and after the fact(s). Again, this is as it should be.
And as a member of TheMotte, a rationalist adjacent site, do you accept responsibility for how the fleas you apparently laid down with tried to firebomb Sam Altman's house?
Yes, a shadow of a shadow of a shadow of a shadow of the responsibility falls on me. I don't really worry about it, because at that far of a remove it's just the cost of being alive in society.
I'll believe you are sincere in your "personal responsibility for other people being bad" stance when I see you apply it to yourself and accept personal responsibility for bad people existing in your own groups.
I'm not writing an autobiography for you, so I guess you'll never believe me. Darn.
ulyssessword
12d ago
There were reports, but the messaging is currently leaning the other way.
Journalist(?): "One Service agent told me the shooter is confirmed dead."
Trump: "The shooter has been apprehended"
ulyssessword
12d ago
I'm a huge critic of the "small number of people did something bad so everyone tangentially related is responsible or guilty" sort of arguments,
I'm a critic of "My movement is only the good people, and the bad ones are unrelated." Sorry, but if you lie down with dogs, you wake up with fleas. Your stance would give zero consequences for extremism.
What does it matter if the guy who shot a police station was actually a boogaloo boy false flagging instead of a BLM protestor? No one is accountable for his actions except for him.
As an example, imagine that there was a gas attack on a public place. The police arrest the perpetrators, and discover that they were all members of the California Chemistry Club. Shortly thereafter, there's another gas attack and the perpetrators are also members of the CCC. And again, and again, and again. It's weird that it keeps happening, but it's not like the Club has any relation to the attacks. No one is accountable for the the perpetrators' actions except themselves. Under your framing, people couldn't even think that the organization might be promoting or benefiting from those actions, because only a small number of their members are carrying out attacks.
If Boogaloo Boys are shooting up police stations, then it's evidence that they're a violent group and should be (formally or informally) punished for that. The alternative is playing whack-a-mole after the fact.
No bad tactics, only bad targets, maybe? That's pretty much #2
More options
Context Copy link