That’s a bit of a canary is it not? You don’t need to say that and wouldn’t know to say that unless you had worked in the space or wanted us to think you did :)
I don't have links. I'm a yeshiva student and many of my friends study in Israel and/or fly back and forth and I know multiple people who used polymarket to make flying decisions.
> questioning the fundamental basis of this.
Empirically, you can look at https://calibration.city/ (among other such trackers) look at polymarket, filter by market midpoint and you'll see that if a market resolves in a year, and 6 months in it's at 30%, the actual event happens at (remarkably close) to 30% of the time.
Theoretically, it relies on standard market theories, like efficient markets hypothesis etc. Basically, however corn comes ot be valued correctly, much of the same mechanisms are present here
Doesn’t Orthodox Judaism (like all religions) look quite harshly upon all forms of gambling? How is Polymarket kosher?
To be clear, I didn’t question efficient market hypotheses - my stance has been pretty clear along the thread, questioning the value of the kind of information gambled upon in popular prediction markets.
Yes, it does. I never gambled on Polymarket, I look at it to figure out the odds of things I care about.
I thought I explained the value quite clearly. ~10 years ago if you wanted to know the odds your flight will get canceled due to war, you had to trust the hyperventilating talking head on your favorite cable news channel (who's job it was to keep you watching...). Now you get basically the actual odds. If that's not value, I don't know what is.
Right. And I don't think the abuse of the vetting people is by accident. I think it's a vulnerability, where people in positions of "collecting dirt" on others, often end up fabricating the dirt, and doing other very bad things because the power imbalance of asymmetric information corrupts.
COme to think of it, maybe that's why priests who take confessions are also correlated with abuse. Something about having this assymetry over many others maybe scrambles their moral circuitry...The Catholic conneciton is just a theory that surfaced now tho, haven't thought it more than that. But the badness of the vetting people is certain. Sad that governments have to tarnish their good names employing such miscreants.
There's an old interview on C-SPAN's BookTV with a CIA polygrapher. He seems to genuinely believe in the validity of the polygraph, but watching the interview, I was convinced that the only value comes from intimidation and stress.
(all-caps bad transcription)
> THE ESSENCE OF A POLYGRAPH TEST IS IF YOU HAVE SOMETHING TO LOSE BY FAILING A POLYGRAPH TEST IF YOU WILL, OR SOMETHING TO GAIN BY PASSING IT, THAT IS WHAT MAKES THE POLYGRAPH EFFECTIVE. WITHOUT THE FEAR OF DETECTION IT IN A SIMPLE WAY AS I CAN PUT IT THAT IS WHAT MAKES IT WORK. YOU HAVE TO BE AFRAID. IF YOU HAVE NOTHING TO LOSE BY TAKING THE POLYGRAPH TEST THAN THE PRESSURE IS NOT ON YOU. BUT AS I SAID THAT IS WHAT MAKES YOU WORK. IT HAS TO BE PROTECTION MORE THAN GILTS. NOW YOU MAY FEEL GUILTY, BUT FEAR OF DETECTION IS THE OVERRIDING CONCERN IN IN A POLYGRAPH TEST
That's the point though. The testers wouldn't actually abuse their victims without the conviction of doing something righteous. Or they would, accidentally or intentionally, spill the secrets.
But if you make even the instruction material lie, then there is nothing that could be leaked and "expose" the system.
Maybe Reformation religions require belief, but the paganism was a set of rituals known to work (by virtue of having worked before), sort of a like a spiritual experimental science. Belief was not required.
Religions don't necessarily work because people believe in it, either. There are a number of religious sects that started with end of the world prophecies.
I think that religions work the opposite way: people believe in them because they work. Since the purpose of religion is generally to explain the nature of reality and how to flourish in it, it needs to work for you. If it doesn't, you either just go through the motions, or quit and find a different religion (or swear off religion, which is sort of the same thing).
Reminds me of Julius Caesar describing the druids. Part of his political career meant precisely performing important orthopraxy. He probably didn’t meet a druid, but amazingly described them playing the same role he did as Pontifex Maximus.
The orthopraxy requiring those precision rituals, take Rome and Greece, had little or maybe no mandatory beliefs. City-state-sized gods in Mesopotamia probably functioned the same way. Traditions still have precise orthopraxy today. But we talk about differences in belief whereas Caesar doesn’t even acknowledge any.
Charitable read, would suggest slight touch of tongue in a cheek.
A bit of spelling it out
Point-1. People just interpreted that paganism works.
E.g. Somebody made offering to gods, and year later won a war - proof.
Point-2 paganism had this transactional notion with gods giving and taking based on your offerings.
While christianity on the other hand does not promise anything good in this life (the only promise being: bear all the bad things in this life, you will be rewarded in the afterlife), so there can’t be proof.
Exactly, the whole point is to put someone into an interrogation scenario for hours or days, where you control whether nor not they "passed". Unfortunately, it probably has zero effect on psychopaths.
>I thought it's settled science that polygraphs don't work
Of course they do. And if you read the article in the OP you also realize why.
Polygraphs are an interrogation tactic, you can force a subject into a somewhat ridiculous procedure and ask them threatening questions, creating an disorientating situation. Afterwards you can accuse them of having "proven" that they are a liar. Polygraphs work, it just does not matter whether the machine is on or off.
Training on the CoT itself is pretty dubious since it's reward hacked to some degree (as evident from e.g. GLM-4.7 which tried pulling that with 3.0 Pro, and ended up repeating Model Armor injections without really understanding/following them). In any case they aren't trying to hide it particularly hard.
My guess they mean Google create those summaries via tool use and not trying to filter actual chain of thoughts on API level or return errors if model start leaking it.
If you work with big contexts in AI Studio (like 600,000-900,000 tokens) it sometimes just breaks downs on its own and starts returning raw cot without any prompt hacking whatsoever.
I believe if you intentionally try to expose it that would be pretty easy to achieve.
One thing; you're supposed to write "Cannot confirm or deny my affiliation with the CIA"
reply