ThenElection
No bio...
User ID: 622
Of the OT-eligible workers who are making substantial bank from overtime, how many of them are actually Trump voters? Genuine question: of the couple people I personally know who make $50k+ in OT, they are all people working for the city. (Though I suppose many in that situation are cops and firefighters, who probably break a different way than most public sector workers.)
I can understand where the motivation comes from for charging him. Removing the political aspect, if there's a bar fight, as a society we don't want the altercation to progress to shooting. And morally in the bar fight situation, I would think it's inappropriate to escalate to shooting, even if you're being tackled to the cement and the other person instigated it. Not that I have any real sympathy for the person shot: seems to be a FAFO situation.
What gets me is the pro-Palestinian not also getting charged. They seem basically equally at fault, engaging in potentially deadly escalations. Give both a misdemeanor for disturbing the public peace.
I think it's less blue-collar vs white-collar and more waged vs salaried. And the professions that are most aligned against Trump are disproportionately salaried (I think?) My guess is that there's a big split in voting patterns between salaried and non-salaried workers, perhaps even more than white-collar vs blue-collar.
Some books are fun. The key is ignoring people who say you have to spend all your time reading neurotic navel-gazing about the Female Experience if you want to be a good person. If something is annoying or boring... You just drop it.
My brain works by splitting it into straw and berry, and then adding 1 and 2. As for how I know how many rs are in each,"straw" just seems obvious, similar to if someone asked me if 9 is bigger than 8. "Berry" is a bit more complicated; I think my mind goes "it has an r in it, but it's a weird word, so it's probably 2."
That's basically half of it.
The other half is using the good responses as a signal in RL on the model. An interesting comparison would be vanilla 4o with the built-in CoT techniques and the RLed model.
One interesting thing is for the hidden thoughts, it appears they turn off the user preferences, safety, etc, and they're only applied to the user-visible response. So o1 can think all kinds of evil thoughts and use it to improve reasoning and predictions, so long as they're not exposed explicitly in a way that would Harm the end user.
They're doing a gradual rollout; supposedly all premium users should be getting it by EOD.
Been playing with it for the past hour. Subjectively, it seems a small but meaningful step better than Sonnet 3.5. But that understates the importance of it: we're not getting o1, but o1-preview, which is substantially inferior according to the posted benchmarks. And the amount of inference-time compute being used is pretty limited, and OAI is seeing capabilities scale through orders of magnitude more compute.
Last week, I was thinking that maybe we had hit some LLM-plateau. Now, I don't.
Back in 2016, Trump faced the same moderator hostility and still managed a better performance.
The dominant strategy is to come up with a couple of well-rehearsed answers and pattern match questions to them. Even someone really sharp and articulate can at best match that performance. When there's around 5 minutes per topic (all of which are obvious beforehand), there's no reason not to.
Best case is to put some traps in the mini speeches to goad your opponent into going off script; a disciplined opponent knows to ignore them.
Polymarket not impressed with Trump's performance so far:
He's not selling himself to anyone not hugely online. Kamala is giving a serviceable performance, sticking to canned speeches and mostly ignoring the questions. Trump is, at best, picking fights on topics that aren't his strengths and doesn't seem to have a coherent strategy.
No one cares about Viktor Orban's endorsement.
Transgender surgeries on illegal migrants in prison.
Can't say I'm too impressed by Harris, but she's hitting her points much better than Trump.
setting it to the ideal temperature
What is this, though? I agree that we should strive for the capability to set the global thermostat to whatever we want, but there are genuinely diverging interests here. Maybe Burkina Faso wants a year round balmy weather for its tourist industry; maybe Muscovites want to wear shorts in January.
It's still better to have that control than not, and probably there's some clever market design where countries can bid to set the thermostat.
There's more to AI than chatbots. GDM had its IMO proof solver and AlphaProteo, which, in normal times, would be potential candidates for technology of the year. The only reason they're not is that they are somewhat obvious next steps in existing research programs, coupled with lots of engineer time and compute.
I do agree with you that we may be in the plateau of the transformer S curve; we'll know more when OAI releases its next model. In the coming weeks, as they say.
-
Winning is winning. I think people here overstate the level of committed ideology among practicing politicians. They mostly want to win and be celebrated by culture.
-
There's the time honored strategy of campaigning one way and then governing another. You'll have less public support to implement your maximalist goals, but you'll also have won an election (and helped more downballot Democrats win their elections). That leaves you in a better position to achieve maximalist goals than losing and being the minority party. When you wield power is when you try to shift public opinion: you have more tools at your disposal.
I kind of feel bad, because I shared the article merely because of the silly headline. Apologies for making you read it.
It's extremely hard to find anything (at least in English language media) that represents the position of proponents of the wall. Presumably there are some: a majority of DR citizens support it. But most articles are just about how it's racist and inhumane and escalating tensions.
For more information, there's Deconstructing Border Walls: Tackling the Dominican Republic’s anti-Haitian racism.
Well, the previous messaging is baked in already. But although the best time to have good messaging is yesterday, the second best time is today.
There's a nice side benefit: Republicans will then say "she's a weather vane who's abandoned all her previous policies!" That does some damage to her, of course, but it's mitigated because voters hear "she abandoned a bunch of failed policies and is more moderate nowadays."
I think the best line of attack would be portraying Trump as a buffoon who lacks the work ethic or principles to accomplish anything or even do his job. Not a neo-Hitler or American ayatollah, but a huckster whose entire vision is driven by whatever talking head he last saw on Fox said. Mention how many games of golf he played while President (ideally, claim he has a terrible handicap and draw him into a prolonged argument about how good he really is). Have a long list of his broken promises and, if possible, at least some plausible sketch of a story of how the Biden/Harris administration actually fulfilled them. I'd encourage her to heavily embellish those stories; if there's even a remote kernel of reality to them, she won't get any flak for it, and even if it's an outright fabrication it doesn't matter too much. At the same time, represent herself as a competent workhorse who's capable of handling the job of President. Have defenses at the ready for attacks around her being too liberal, and feel free to jettison or reject any policies that are inconvenient.
Predictions: the debate will be boring; Kamala will say a bunch of vague things and throw some zingers at Trump; Trump will offer some stream of consciousness responses and some Kamala jabs; neither will have a collapse a la Biden; hand-picked focus groups will say Kamala won in a blowout; nothing really happens in the polls.
I'd love to see AI come up in the debate, though.
Harris is a better candidate than Hillary Clinton. Clinton treated voters with apathy at best and unveiled contempt at worst. Kamala, whatever someone thinks of her substance, is actually trying to appeal to voters.
The schadenfreude is going to be the best part of this election. I very slightly lean Kamala, but it's going to be a wondrous night schadenscrolling Twitter/Reddit/Facebook if Trump wins.
For California in particular, I think more electable Republicans. Less crazy Democrats would be good and probably closer to my actual policy preferences, but having a single party system itself seems to lend itself to bad governance (at least in the context of American politics). Moderates and extremists will have different policies and spar with each other, but they close ranks when there is corruption or something that could affect the reputation of the party as a whole.
These obviously had explosives physically implanted in them. If someone is reporting there were no explosives, they are either 1) an Israeli operative trying to propagandize to make Israelis seem more powerful than they are or 2) someone who can't identify explosives, or someone looking at pagers that weren't modified.
The real risk for supply chain attacks is spy software and backdoors: information is power. And yeah, any country where all its electronics are made or handled by a hostile foreign adversary should be worried.
More options
Context Copy link