12 Comments

The fact that calls get scanned (Echelon program) is already established. The question of whether everything gets stored requires a feasability calculation (which people have come up with in the other comments), but because the black budget of the US government is so large (on the order of $100 billion I believe) we won't know for sure if everything gets stored until such information is declassified, which may not happen anytime soon, you see what Robin Hanson forgot to mention is that stuff doesn't just get declassified automatically after x years, it only happens after the stuff in question has ended and a new wind is blowing in Washington D.C. so if everything is being recorded and it's still being recorded in 2063 then in 2063 the government won't declassify that they recorded everything in 2013.

Expand full comment

The judge of the bet can't be anonymous, but the bettors can be. You can buy/sell predictions using blinded tokens. For example using an Open-Transactions asset over Tor, paying with Bitcoin.

It doesn't exist yet and wouldn't be very stable (most of your bet would be on Bitcoin volatility) but it's technically feasible.

Expand full comment

This seems to me a great place for a prediction market. It seems quite likely that the truth will be revealed within a half century, and if this claim is true hundreds of people must know who might be tempted to make a little extra money via anonymous bets.

Given that no bets can be truly anonymous (as participants' identities must be verified at maturity), the risk of losing one's career for what will probably be no more than a few hundred dollars would be enough to render the market unreliable.

Expand full comment

Hypothetically, how would one go about verifying said fact?

Expand full comment

"This seems to me a great place for a prediction market. It seems quite likely that the truth will be revealed within a half century,"

My recollection from InTrade was that the longer dated contracts were quite illiquid. A prediction market on this would require the better to post margin or collateral for a very long and indeterminable amount of time.

The requirement to commit margin for such a long period of time makes it highly unattractive even if the better does believe she holds an edge. Since they're binary contracts that potentially trigger at any time the participants would have to post collateral equal to 100% or more of the notional value of the contract.

For example if the contract's priced at 53% and I believe the likelihood is 68%, that's $100 of capital I have to commit for likely 50 years or more just to earn $15 in return. That only represents an annualized return of 0.27%.

Now you could say that the prediction market could be very sophisticated about how it manages margins. That it could seamlessly integrate with people's financial portfolio, or even their home value, and accept that as collateral. This is much easier said than done, and even the largest and most sophisticated future exchanges don't make it that easy.

Finally even if all that happened, the collateral requirements and haircuts would mean that investors could only gain small exposure to these "ultra-long" prediction positions. Futures exchanges usually apply a 30% haircut to equities. So even if an investor could consistently find ultra-long bets where he had a 15% edge, it would only enhance his portfolio's return by 0.19% a year.

That hardly seems worth the effort of the research and investment process given the return enhancements. Retail investors would be disinterested because that return enhancements would only represent a few extra thousand a year. Hedge funds would be disinterested because the huge collateral requirements mean they can't gain leverage. And large institutional managers would be disinterested because they usually have quite conservative and restrictive IMAs that prevent them from investing in anything exotic.

If prediction markets have a chance of going mainstream, particularly binary predictions, it will be in short-term bets.

Expand full comment

I've found that it's never wise to underestimate the NSA's ability to do anything with data. There's a reason they're said to measure their computers by the ACRE rather than conventional measures.

Expand full comment

Added May 9. OK, then here's my best speculation: Robin hasn't found a way to get predictions where there's no unequivocal criterion, and he hopes someone will suggest one. He'll be disappointed. A severe limitation on using prediction markets to determine likelihoods is that there must be an unequivocal criterion. In ordinary markets, the inevitable ambiguities in many contracts is resolved using the expedient that the contract says what the court interprets it to say. A similar expedient exists for predicting economic trends: a recession, for example, is when the appropriate bureau says there's one. Either the criterion must unequivocally indicate independent variable OR how some official body views the independent variable's status must be important regardless of whether it indicates anything about the ostensible underlying variable. Someone may claim Gore really beat Bush, but no one will deny that the fact that Bush was proclaimed the winner was important--more important than who "really" won.

For conspiracies, we want the real truth, not a surrogate. Prediction markets can't help us adjust our "priors" unless there's a way to measure some important result without equivocation or doubt. From a Bayesian standpoint, these events are truly rare: for most things, there will seldom be unequivocal certainty. Prediction markets are limited to that rare subset of events.

___________________I don't understand why Robin doesn't respond to the obvious question repeatedly voiced: what's the criterion, where every conspiracy theory continues to have believers and detractors notwithstanding the evidence?

Expand full comment

It seems quite likely that the truth will be revealed within a half century, and if this claim is true hundreds of people must know who might be tempted to make a little extra money via anonymous bets.Given that we don't have good knowledge now about Echoleons capabilities in the early 1960s what makes you think that it will be easier to know the current surveillance capabilities in a few decades?

Why do you believe that the NSA will change to be more transparent about it's surveillance capabilities?

If you listen to the interview, it's also useful to note that Tim Clemente doesn't say that the NSA directly has the information. He speaks about oversea assets.

That sounds like Echealon where according to rumours the UK wiretapps all of the US and the US wiretapps the UK. Afterwards they swap information. That's the legal loophole. The US has no law against aiding the UK to wiretappp US conversations and vice versa.

Expand full comment

Regardless of whether every call gets stored, we know for a fact every call is scanned for certain words, including the calls of foreign citizens in foreign countries and many innocent conversations are indeed recorded without judicial oversight. No doubt some of the information obtained in this way ends up in the hands of corporations too. All in the "land of the free"...

Expand full comment

They didn't keep it successfully a secret. Well informed people had some evidence that this is the case before Tim Clemente admitted to it.

Andy Müller-Maguhn who just published his book with Julian Assange did a talk ( http://www.youtube.com/watc... ) to announce that this is the case.

Someone did give Wikileaks all the pager messages sent on 9/11 years after the fact. This means that this person has to have access to a database that contains those messages.

The fact that someone did store those messages even before the Patriotic Act was passed, clearly demonstrates that the average person doesn't have a clue who stores what.

Just because you are not willing to accept something and mainstream authorities at places like the New York Times aren't willing to accept something doesn't mean that it's a real secret.

When it comes to keeping government programs secret, the Mannhatten project was successfully kept secret for years dispite employing over 100,000 people.

When it comes to us knowing how the US government used National Security Letters our knowledge comes from less than a handful of people standing up dispite 100,000s of letters sent. The story of one of those people Nicholas Merrill illustrates how hard it is to challenge this kind of secrecy ( https://www.youtube.com/wat... ).

Expand full comment

It'd be pretty cheap actually. Somewhere around $50-100mm/year. That's to store one year's worth of domestic phone calls at 8kb, if memory serves.

Not sure how big Amazon's data center(s) are, but this needn't be particularly huge, I suspect.

Utah data center may permit higher quality storage, and also allow for storage for decades to come.

Edit:

This guy thinks it storage costs may be as low as $500,000 for one year's storage.

http://test.neilk.net/blog/...

Expand full comment

I feel like its pretty easy to debunk this, consider the amount of data this would generate, the amount of storage you would need, and the cost and power requirements for all this, then consider how difficult it would be to keep it a secrete. Also consider the cost of memory at the time it was completed, and the program begun, not the current cost.

Our government is not operating a cloud storage service bigger than Amazons.

Expand full comment