I'm not sure providers would shoot themselves in the foot by not allowing you to buy more usage. It seems like both OAI and Anthropic could use all the subs they can get.
Personally I’m a fan of the non sports side of these apps, Polymarket in particular. They are putting a financial incentive behind accurately predicting the future, which makes these apps basically the most accurate predictor of the future. Available to anyone to view, for free.
It makes sense why they went towards sports gambling, because the market is so big.
I think this is kind of a naive take. There's a reason why you can't take a license insurance policy out on a stranger. Being able to bet that there will be a wild fire or if a plane will be late creates terrible incentives. It's also a new way to do insider trading.
Most uncertain things that people need to hedge (currency, commodity prices) already had a hedge.
In practical terms can someone really give me an example of how they benefited from knowing the future (materially, not emotionally) from a prediction they got from Kalshi? I am honestly curious.
The only one I can think of is, having a better idea of which way the presidential election goes might've been useful for some people, or some other large election. There's a lot of egg on the faces of all these mainstream media polls, which still continues to trigger some people.
So what happens when someone like musk spends a few hundred million to skew what the probability looks like? Suddenly people see that a politician has a 90% chance of winning, and they feel silly voting for the operation?
These are not polling places, they are places for capital to influence odds.
Eh, sounds like the excuse I kept hearing as to why Trump had favorable odds in the last election. Supposedly a foreigner trying to influence our elections, totally not the Democrats kinda not having a candidate. It would be a very expensive (on average) way to probably not make a difference. Cheaper to just buy up some media companies, which do the same thing, saying their preferred candidate is going to win.
Small elections are susceptible to corruption this way, though.
There’s been plenty of blatant voter interference, and billions spent influencing American voters, not really seeing where your point is based. More handles to steer democracy with money is not better, period.
Most uncertain things that people need to hedge (currency, commodity prices) already had a hedge.
That's way too narrow of the categories. Here are some more cases, and you tell me where can I hedge it? And don't tell me the answer is "go to Goldman Sachs and they create any kind of derivatives to hedge for you", that only works if you are a huge enterprise. I'm talking about SME, or even one person company with $1-5M ARR.
(By the way, I do trade on Polymarket / Kalshi, had a job interview with Kalshi many years ago but did not take their offer since I disagreed a lot with their product design. So all of this is my own opinion).
1. Suppose I run an ice cream shop in midtown Manhattan, and my foot traffic depends on how good the weather is. If it's sunny and dry, I expect a lot of traffic, vice versa, if it's thunderstorm, hurricane or snow, I would run the day without any revenue (and a lot of ice cream waste). What is the way for me to hedge weather risk that could seriously affect my business?
2. Suppose I'm a restaurant owner in Koreatown in LA, and during World Cup I run a restaurant expecting a lot of Korean coming to my area (and my restaurant). So I would ramp up my business, hiring a lot of staffs, sourcing a lot of ingredients from the vendors. Then the biggest risk (and anticipation) is that Korea would be knocked out very early, which unfortunately that precisely happened. So the foot traffic evaporated, and with that the revenue. How could I hedge that risk?
3. Suppose I'm a distributor / reseller who needs to import goods and products from other countries and resells it back to US for profit. Barring the import / export regulations, the biggest risk is tariffs: the orange man can (and did) impose 50% tariffs from China, and all the sudden I face a severe supply-demand imbalance with a lot of contract cancellation and such. How can I hedge such tariff risk from the orange man? (Obviously, the alternative is that I wish we could have elected a more competent leader, but at least 75 million American disagreed with me).
> Most uncertain things that people need to hedge (currency, commodity prices) already had a hedge
What happens when the people with pitch forks come for your derivatives and call it gambling? They did during the 2008 financial crisis. It sounded pretty much exactly like your take here, one abstraction jump away.
Then after they've come for your derivatives, they come for your banks. Because what is a bank except skimming off the top of the company doing the real business?
Then they come for your company. Because what is a companies except skimming off the top of the workers doing the real business?
And then you're standing in that bread queue wondering how it happened, even if it was all 100% predictable.
ah yes the slippery slope that leads from banning prediction markets that didn't exist 10 years ago to banning erm, companies. definitely a guaranteed outcome you should make a bet.
> They are putting a financial incentive behind accurately predicting the future,
Unfortunately they also place a larger financial inventive on influencing the outcome of future events. Or incentivizing corruption within the inner circles of these events. Because the larger financial upside in these transactions is to bet against what the market thinks, and influence the outcome.
I don't think that is logically correct. Commonly perceived probability has been proven frequently wrong whether naturally or due to manipulation. The reasoning here is not exactly the same, but closely aligns to bandwagon fallacy.
It would be interesting to see how frequently Polymarket predictions bet against the most common prediction and by what margins.
It's the least inaccurate way. There's money to be made if you're sure you have a better way. The real problem is we're already seeing these markets influence the outcome.
You make two valid points, but they still do not qualify the validity of these markets as accurate predictors. There is a difference when people, like Berkshire Hathaway really dig into the financials and management of a company to predict its future versus somebody like me who just money on a stock. In one case an investor seeks to make money and actively influence an outcome, but in my case I am just dropping money and hoping it out paces the rise of inflation because banks are no longer reliable stores of financial security.
Polymarket projections do not provide investigative insight, as in the Berkshire Hathaway example. It's just gambling. Any influence drawn from these predictions is incidental and reactive, as opposed to active market forces like buying/selling.
Nobody is saying it's better than stock market predictions of assets and company financials. Though there are some asset price predictions on there that exactly match the traditional markets, so it's no worse.
> They are putting a financial incentive behind accurately predicting the future, which makes these apps basically the most accurate predictor of the future.
There is plenty of data that shows they actually don't work very well outside of surfacing insider knowledge (which the commercial prediction markets superficially pretend to be against).
For events where you can't have full picture insider knowledge like elections, prediction markets are pretty awful at accurate predictions.
The range of activities in which this effect can be harnessed for societal good is extremely narrow, but I think it's there.
I'd be in favor of a prediction market for which which asteroids are likely to collide with something, and when. Though we'll have to shut it down once it becomes feasible that somebody might go up and nudge one.
Or they're creating a financial incentive for insider trading and manipulating certain events.
I mean look at all the oil market manipulation in the Iran war. The pattern was repeated so many times: short oil, announce a ceasefire deal with imminent on Sunday before the markets open, cash out on Monday, rinse and repeat.
The argument is that we get transparency on what's likely to happen. I don't think there's any evidence of this. I've seen any number of very poor predictions coming from Kalshi and Polymarket. I have also seen some clear manipulation.
> They are putting a financial incentive behind accurately predicting the future, which makes these apps basically the most accurate predictor of the future.
This is so fallacious. Putting money behind something doesn't make a prediction of a non-knowable more accurate.
Tarot card readings by a psychic must be more accurate predictor than a tarot card reading by an amateur because professional tarot card readers have a financial incentive to be right, so people continue coming back to them for advice. Getting a tarot card reading therefore makes this basically most accurate predictor!!
> They are putting a financial incentive behind accurately predicting the future, which makes these apps basically the most accurate predictor of the future. Available to anyone to view, for free.
Here, let me predict a future where someone will die. On the other side, let me pay for a hit on the same person. /s
^It's a bit extreme and exaggerated, but it's like that on purpose to make a point.
Regarding the Pareto frontier and related benchmarks, I have a hard time taking anything seriously that claims that Opus is anywhere near the intelligence of Fable. Are there any benchmarks that haven't just been benchmaxxed that more accurately represent actual usage?
FrontierCode is prob the closest. [1] It's closed source (so no direct benchmaxxing), and it was calibrated by 20+ open source maintainers. It shows Opus 5 (medium), beating out the other reasoning levels by a large margin. i.e. Opus 5 w/ higher reasoning levels actually reduces performance. [2]
However, you'll have to gauge for yourself how closely their tasks resemble your tasks.
Nobody is forcing people to be employees of startups. They choose to do it of their own free will. If they think they can generate immense value from their own ideas they are free to go try to do that. Hiring people to do a job for a wage they agree to isn’t swindling.
Let’s ignore all the political and socio economic dynamics driving the current trends. Squeeze people’s livelihood by rent seeking; pillage public resources while externalizing as much of the risks and cost of doing business back to public; limit access to capital to few networks. Then pretend it’s in people’s power to choose to earn more if they would only work harder.
No one is arguing that founders shouldn’t earn the rewards for their work and risk. But to act like it’s all fair business and freedom of choice is rather glib to the point of being callous.
It’s difficult to determine which individuals are involved and even if you manage to do that they almost certainly live in countries without extradition.
Who doesn't have hard billing caps for inference? Microsoft, Google and AWS my friend. And you know who uses Microsoft, Google and AWS? Almost all big corporations do use them instead of direct OAI or Anthropic API because all their contracts and infra are built around the big cloud providers.
> Long-running tasks like batch mode completions and agent sessions may incur overages beyond your project spend cap.
> Billing data processing times can be delayed in AI Studio, up to around 10 minutes. You may experience overages beyond your project cap if billing data hasn't processed before more charges are accrued.
reply