The CIA and Jeff Bezos Bet $30 Million On Quantum Computing Company

An anonymous reader writes "The CIA's investment fund, In-Q-Tel, and Amazon founder Jeff Bezos have invested $30 million in a Canadian company that claims to build quantum computers, reports Technology Review in a detailed story on why that startup, D-Wave, appears to be attracting serious interest after years of skepticism from experts. A spokesman for In-Q-Tel says that intelligence agencies 'have many complex problems that tax classical computing architecture,' a feeling apparently strong enough to justify a bet on a radically different, and largely unproven, approach to computing."

Comments

D-Wave might actually be legitimate

[AdmiralXyz]

Just a quick FYI: for those of you still assuming that D-Wave is a bunch of snake-oil salesman (like I did for a long time), take a look at this bit from Ars Technica. Basically what they've built is not a genuine quantum computer, but a sort of "quantum optimizer" that delivers speedups for some kinds of problems. Their crime might be that they just use too much marketing hyperbole, instead of being complete frauds.

Re:D-Wave might actually be legitimate

[exomondo]

The accuracy is the fundamental problem, with no error correction (or at least indications that there is an error), which is one of the biggest problems with their approach, it's worse than useless. In the protein folding experiment it got the correct answer just 13 out of 10,000 times.

Re:D-Wave might actually be legitimate

[w_dragon]

That depends on the problem. I assume the CIA wants it for breaking encryption, which means they want it for factoring large numbers. That's a problem that is really hard for a normal computer to do, but really easy for it to verify. If factoring a 1024 bit number takes 10000 tries, and each try takes a second, you're still several orders of magnitude better than the current state of the art and you've rendered many of the current common encryption schemes useless.

Re:D-Wave might actually be legitimate

[khallow]

In the protein folding experiment it got the correct answer just 13 out of 10,000 times.

Getting the right answer once can be good enough. It depends on how the relative cost of checking if an answer is correct. I gather this would be used to figure out NP complete problems (which I might add, the protein folding experiment may not be in) where finding the answer isn't known to be doable in polynomial time, but it can be checked in polynomial time.

Re:D-Wave might actually be legitimate

[Anonymous Coward]

The NSA wants it for decryption and is smart enough to know an adiabatic quantum computer can't be applied to factorization problems. The CIA wants it for the same reason Google did, image comparison.

That's not even pocket change for either of them..

[rsilvergun]

That's like me betting a nickle. Strike that. A plug nickle.

What is it with these public-private partnerships?

[js33]

The CIA's investment fund, In-Q-Tel,

You know, the government has absolutely no business running an investment fund, especially a "secret" one where it looks like there's no meaningful oversight. This is we the people's money, and we the people have no interest in being the angel to some sleazy fly-by-night foreign start-up who just wants to suck at Uncle Sam's ever-so-generous teat.

What does quantum computing mean for developers?

[proca]

If I'm still developing when quantum computing becomes ubiquitous, how will programming work? Will booleans suddenly have 8 states? True, False, KindaTrue, MostlyFalse, Truthiness, TotallyBogus, WayCool, Cowabunga?

Re:What is it with these public-private partnershi

[Charliemopps]

I think its rather obvious that it's not secret... since we're talking about it. I'd rather the CIA be investing in new technologies and improving society. At least they didn't spend $30 million starting a war somewhere.

Re:What does quantum computing mean for developers

[aNonnyMouseCowered]

No need to wait. I have a quantum computer right here in my pocket. It's called a coin. You want eight possible states? Add three more nodes. It's highly efficient for answering life's toughest questions. And if I don't like the answer I can try again.

Re:What is it with these public-private partnershi

[strat]

The short answer is that the times have changed from back when government-funded applied research was a primary source of startup innovations. The reality is that small companies move faster and are more able to adjust to surprises in an agile manner than the Government. Now the tables have turned and the Government needs mechanisms to find new things because it's certainly not inventing them all in-house.

Speaking as one of the other members of the population, I have a few mixed feelings about the government using public funds for equity buys. Conversely, if that mechanism allows the USG to more rapidly gain access to novel inventions than they have and those inventions optimize the Government's performance, it's a drop in the bucket and probably saving the taxpayers a bundle.

If you find Google Earth useful, thank In-Q-Tel. When the startup that produced that technology was financed, only realtors in California had ever heard of it.

(Yes, I'm a little biased. I have been a part of some public-private partnerships that have performed well.)

More $$$

[Patent Lover]

Is the CIA once again hoping some expensive technology will actually allow them to finally get something right?

Re:More $$$

[NEDHead]

Is this supposed to be some sort of refutation? If I listed seven things you screwed up in your life, would that damn you eternally?

My point was simple, but I will restate it here for your edification: Their job is secret; they don't tell us what they do, as a rule; you and I have no basis for judging their performance; they may (or may not) have had many major successes that we would celebrate if we knew of them.

We just don't know.

Re:That's not even pocket change for either of the

[Sulphur]

Sure wish Bezos would spend a nickle to make Amazon search actually work.

Is 13 out of 10000 tries good enough?

Re:What does quantum computing mean for developers

[Anonymous Coward]

X*(Y*2/Z-1) finishing date, but managers typically misread it as X*(Y*2/2-1), which is why deadlines are always set impossibly soon. Just to clarify, X is ten years from any starting date (3650 days). Delivered Z is antecedent penalty; a reciprocal of the sum of all previous & related technologies squared 1/((q1+q2+q3...)^2). That's why it can take millions of dollars to shorten development time by mere days; the fancy equipment budget negates the penalty of the antecedent technologies. Reinventing the wheel is when a budget of under $20 results in a "yesterday" finishing time, which is a symptom of a project delivering last-age tech (stone,bronze,steel,etc.). New tools and science push the next-age, which means gradually eroding the antecedents impact. Because of how many underlying projects led to current ones this-age, there's currently a very heavy price to get anything done quickly.

Better tools and scientific knowledge need to become ubiquitous, or else the cost of pushing technology forward grows ridiculous, it's burden becomes too high and new advancement stagnates.