The smartest guys in the room


I was reading about this guy, Vitalik Buterin, who is supposed to be some super-smart crypto guy. There’s an infestation of them in the dudebro community.

Buterin comes from a long tradition of Silicon Valley special smart boys, who have had it hammered into them that domain expertise — i.e., actually knowing stuff — pales into insignificance compared to pulling ideas out of your backside by virtue of your superior intelligence and upbringing and social position.

He was taught this by other Silicon Valley special smart boys. Peter Thiel literally paid Buterin not to go to college any more, based on this theory — that one special smart boy reasoning from first principles will surely beat the accumulated experience and wisdom of mere humanity.

There are a million of these guys, and they all have long and wordy blogs.

Ooh, that’s mean. But if you’re like me, you want to know you can know how out of touch with reality they might be. Maybe they actually are really smart guys, you know. But here’s an example of thinking out of the box. Get a good grip on your jaw before it rolls under the table.

Before founding Ethereum, Buterin put considerable effort in 2013 into trying to convince investors to fund him in constructing a quantum computer. (Note that no quantum computers able to solve practical problems are verified as existing as of early 2017.) His plan was to use this quantum computer to solve computationally infeasible problems that can’t be done practically on an ordinary computer, such as reversing cryptographic hash functions.

Since he didn’t know how to build a quantum computer, his plan was to simulate one on an ordinary computer – since this apparently wouldn’t count as just running a program to solve the impossible problem. This was an idea that had long been put forward by Jordan Ash, his associate in this endeavour, who had put considerable effort into this startlingly crank mathematical notion.

How do you reveal that you know nothing about ordinary computing and nothing about quantum computing at the same time? I wonder if Peter Thiel would reward him with a big grant for this idea.

Comments

  1. says

    :-D

    My intended lesson with these posts was not that Buterin is unintelligent, ‘cos he isn’t. It’s that smart guys who forget that as humans they are dumbasses are gonna do some dumb stuff.

  2. whywhywhy says

    Let’s just say that one could simulate a quantum computer using regular computing (which you can’t do). My gut (dudebro for the ultimate truth) tells me it would be less efficient than simply using a regular computer for the task. (Unlike a true dudebro, I am open to being corrected here.)

  3. billseymour says

    Just for fun, I once wrote a PDP-8 simulator that ran on a PDP-8.  (Yes, really; and I still have the punched paper tape to prove it.)

    Simulators always run a lot more slowly than the real thing, and speed is what’s needed.  The issue with the kind of problems he’s talking about is that, with any computer we can even imagine, solving them will take more seconds than there are protons in the universe.  (OK, I’m exagerating a bit; but that’s the basic idea.)

  4. bcw bcw says

    There is a fair amount of work on simulating quantum computers, the purpose is to understand how different computing approaches work and to characterize how errors and error correcting schemes are likely to behave. This is similar to software programs simulating computer and FPGA chips. They are useful diagnostic and learning tools but useless for actual real problems as they are tens to thousands of times slower than a real device. It separates the “software” aspects of the problem from the effects of actual device characteristics.

  5. PaulBC says

    Oops, David Gerard may not have intended it, but now my interest is piqued.

    Buterin is an interesting writer, and his essays are amazing science fictional ideas

    Having interesting if ultimately nonsensical ideas ain’t nothing. Most of the “silicon smart boy” genre strikes me as excruciatingly boring. Does James Damore count as one?

  6. bcw bcw says

    @4 but to write a simulation you have to know enough about the hardware you’re developing to write code to simulate it’s expected behavior. Buterin was focusing on where there was lots of money interest combined with a subject complicated enough that he could snow investors easily.

  7. PaulBC says

    I suppose you could simulate a quantum computer if you were willing to accept exponential slowdown. I’m not entirely sure what it would tell you, but I can see it as part of a CAD tool suite for testing proof of concept on small instances. Long ago, though I forget what the package was called, I worked with an analog circuit simulator. It’s not a totally crackpot idea.

    Of course, if you thought you could replace the hardware in practice with a simulation, you’re kind of missing the point.

  8. chrislawson says

    “The more he talked of his honor the faster we counted our spoons.” –Ralph Waldo Emerson.

    I feel the same way when people talk gushingly of their own intelligence.

  9. Rob Grigjanis says

    I imagine you could simulate a quantum computer classically. A system of n qubits would be represented by a complex-valued vector with N components, where N = 2^n. Then any operation on that vector (the calculation steps) can be represented by NxN matrices multiplying that vector. As PaulBC notes, things get nastily exponential very quickly, sort of defeating the purpose of the exercise.

  10. Matt G says

    Okay, that was a REAL LOL moment! Unbe-freakin-lievable! How can we earth bound biologists compete with this level of genius?

  11. Snarki, child of Loki says

    Since any omniscient deity would (by the simple act of ‘observation’) cause the quantum wavefunction of a quantum computer to collapse and thereby crash the program, any operational quantum computer can be taken as evidence of the ABSENCE of an omniscient deity.

    Microsoft has clearly been working along these lines, so that they can declare that all of their BSODs are ‘Acts of God’.

  12. unclefrogy says

    if you could make a simulated quantum computer on an ordinary modern computer then you would have a way to make a simulated result that you could ‘use”.
    make an imaginary computer to get imaginary answer to an impossible problem of making a profit off of imaginary money forever.
    “step right this way to the egress”

  13. brucej says

    PaulBC @7

    SPICE works (and is very widely used) because the theory and practice behind analog circuits electronic circuits are known, thoroughly. That’s why EE’s use it to confirm IC and circuit designs before committing them to silicon.

    That’s not even remotely true for quantum computing.

    What the techbros like Thiel and Buterin are good at is sounding smart to investors who have more money than sense.

  14. birgerjohansson says

    In Britain, these dumbasses are usually found in the current government. I could tell stories, but PZ does not have that many remaining years even if he lives as long as the former queen.

  15. seachange says

    I was living in Silicon Valley when every edge was the cutting edge. Actually learning stuff in college is detrimental because most VCs don’t know what they’re investing in. Thiel is not a dummy.

    The people who won and the corporations that won and the apps and programs that won in Silicon Valley had very little to do with whether or not the products were good, they just had to be adequate. Financing is always the issue. Once financing was secured you could glue onto/kludge another layer/change bugs into features. For many of the silicon valley boys then, they don’t need to talk science, you need to talk Venture Capitalist. They weren’t even silicon valley boys or even Californian. Most of them smelled money and moved right in on spec, which is one of the reasons real estate is so expensive there and in San Francisco.

    It was and still kinda is, a gold rush. Vaporware was totally a thing, just like some strikes of gold veins absolutely weren’t. And sometimes still is for both. This is what this guy sounds like exactly to me, he speaks VC fluently.

  16. keinsignal says

    I’ve written rather simple programs that simulate QC-like processes via wave collapse algorithms. It’s not too difficult to write (especially if you’re basically copying someone else’s code ahem), but it doesn’t perform anything like the real deal would, where the results of that collapse would be instantaneous instead of having to be calculated one bit at a time, often taking hours to crunch through the numbers.

    So, you absolutely can emulate a quantum computer on a classical one, to the extent that it will run QC-like algorithms on simulated qubits and produce answers something like a real QC might spit out. But the emulator will lack the underlying structure & capabilities that make QC special. Benefit of the doubt, maybe the reporter is misunderstanding, or writing unclearly… Maybe the intent was to start learning QC techniques & testing out algorithms without having to actually build the machine or tie up expensive resources even if they did have access to one. That’s what existing QC simulators are for, anyway.

    If this guy really thought his emulated QC would be a way of bypassing the problems of actually building a real one though, just, wow. Erlend Meyer @11 already nabbed the only analogy I can think of – installing MS Flight Simulator on your PC to “solve the problem” of moving your physical body somewhere across the continent.

  17. gijoel says

    Kent resembles Cerebus from Church and State more each day. “God demands you give me all your money.”

  18. snarkrates says

    As a wise researcher once said: “6 months in the lab can save you an hour in the library.”

  19. PaulBC says

    Simulating the colonization of Mars could have some real cost-saving advantages. The high-tech version would require advances in VR technology to be really convincing, but I’m thinking of something simpler.

    You just excavate an underground bunker anywhere convenient and convince the occupants (maybe through post-hypnotic suggestion*) that they’re on Mars. They are told they need to stay beneath the surface at all times for radiation shielding, but it’s still really cool because… Mars! We know Elon Musk’s company has the capability to dig tunnels even if they’re not practical for public transportation. So all the pieces are in place and it would obviously be less costly than really putting people on Mars while providing roughly the same benefit.

    *I’m skeptical that this would work well enough, but as long as it works on most Mars “travelers” you might just have to deal with the holdouts by restarting the simulation and replacing them with more susceptible subjects until it is moving along flawlessly.

  20. PaulBC says

    John Morales@22 Yes, I thought of that after posting. Maybe if you keep the ceilings low enough so no jumping, they won’t notice.

  21. unclefrogy says

    long term space flight simulations have been done many times and I am sure will be done in the foreseeable future. All participants know in advance there is no attempts to obscure the fact of the simulation experiment though not all of the areas being tested are always disclosed beforehand

  22. Allison says

    I believe the book about the Enron fiasco/scam was also called “the smartest guys in the room.” I’m guessing that using that title for your post was intentional.

  23. says

    Buterin was focusing on where there was lots of money interest combined with a subject complicated enough that he could snow investors easily.

    And when that didn’t work, he went over to cryptocurrency instead. Yeah, that makes me really confident in crypto. Not.

    Since any omniscient deity would (by the simple act of ‘observation’) cause the quantum wavefunction of a quantum computer to collapse and thereby crash the program, any operational quantum computer can be taken as evidence of the ABSENCE of an omniscient deity.

    Yabbut what if God doesn’t look at it?

  24. says

    Till I was about 20, I felt like I was always the smartest person in the room. Then I enrolled at Caltech and got to discover what it felt like to never be the smartest person in the room. That was an important part of my education, and it has served me well.

  25. says

    …any operational quantum computer can be taken as evidence of the ABSENCE of an omniscient deity.

    So if a quantum computer doesn’t work, the makers/owners can say it’s not their fault, AND they should get a Nobel Prize for proving the existence of God.

  26. jo1storm says

    You can simulate quantum computers and circuits, it is just they are really slow, especially compared to what would be real deal. In fact, you could make your own simulated circuit if you want: https://quantumai.google/cirq/start/install

    There are even courses in quantum programming, for that moment when the real deal becomes available. But to actually make a proper simulation would take decade of research, not a “genius” in a garage, which is what google and IBM actually did.

    Speaking of quantum I think you might like this comic: https://www.smbc-comics.com/comic/quantum-weirdness

  27. John Morales says

    jo1storm:

    You can simulate quantum computers and circuits, it is just they are really slow, especially compared to what would be real deal.

    Actually, one can simulate what current theory holds to be true regarding quantum computers and circuits.

    But, you know… the taste of the pudding.

  28. nikolai says

    John Morales@22: You could still expect to fool an unfortunate number of people without having to worry about gravity. Britain had a reality TV show where contestants were told (as part of an elaborate prank) that they were on a space station (which was really a set in Suffolk), and the reason they continued to feel the effects of gravity was because they were still “close to Earth”. They were apparently fooled for the duration.

    They called it “Space Cadets”, IIRC, and it aired in 2005 or thereabouts.

  29. PaulBC says

    nikolai@32 I had also considered that space-opera science fiction assumes some way to generate gravity, even though it’s impossible except simulating it with a centrifuge. So you could just tell the participants “Gravity is artificially set to earth level to keep your bones healthy.” or similar.

    I don’t really think anyone is that gullible, but then maybe I’m the one who’s naive. If you simulated the entire trip to Mars, you could skip any dubious form of mind-control.

    You have a lot of room in the budget without getting even close to the cost of a real trip to Mars.

  30. Rob Grigjanis says

    PaulBC @33:

    I had also considered that space-opera science fiction assumes some way to generate gravity, even though it’s impossible except simulating it with a centrifuge.

    Not impossible; since we are diamagnetic, magnetic fields could be used to levitate us, or equivalently, provide artificial gravity.

  31. cheerfulcharlie says

    Simulating a Mars colony? Remember the Biosphere? An experimantal sealed dome meant to test the ability for a self sustaining Mars colony enviroment? It turned out to be a failure. Real world experiments like this are better than simulations. Simulations can have fudge factors. Real hardware, not so much.

    Yoohoo, Earth to Elon Musk! Earth to Elon Musk!

  32. blf says

    anthonybarcellos@27, Basically ditto — albeit before I attended Caltech, I did attend an NSF-sponsored summer science course (in Mathematics), and very quickly learned that lesson.

    On a tangent… People here might be confusing simuations with emulations. A emulator tries to do the same thing as the real thing in the same way, whereas a simulator tries to obtain the same results as the real thing, but not necessarily in the same way (albeit constrained by requiring the same input as the real thing). As least when it comes to microelectronics-design, emulators are usually a sea-of-gates (FPGA) hardware implementation, possibly derived from the (micro?-)architectural models, whereas simulators are software (a considerable mix of hand-coded and architectural models). Just to confuse things, there are architectural simulators, which do indeed try to do in software exactly what the real things does.

    Speed-wise, emulators are the fastest and architectural simulators the slowest. From the last project I was involved with (many years ago now!), a 400 MIPS pipelined dual I-stream (“microthreaded”) (potentially multi-)core system, the emulator ran at something like 4 MIPS (broadly, 4 millions emulated instructions per second), the simulator at about 100 KIPS (one hundred thousand simulated instructions per second), and the architectural simulator at 0.25 IPS (4 seconds per architecturally-simulated instruction).

    None of this means the claim(s) by the quoted fool in the OP is valid, it’s a nonsense.

Leave a Reply