The skepticism is not about fusion power, it's about cold fusion specifically, as well role of proprietary IP and patents in scientific research.
Tokamak[1] (and other magnetic confinement) reactors seem much more promising and have seen slow but non-zero improvements for decades[2]. Some reactors like JT-60 have come close to being energy positive[3] but are not yet self sustaining. Large ongoing projects like like ITER[4] have the potential to break past limits and provide invaluable data. There are also advances in plasma physics which improve our ability to run computer simulations[5] of Tokamaks and suggest new hypotheses for how they can be improved[6].
If Google were serious about fusion, they would invest in this kind of mainstream research. Unfortunately, that's too big for them to own end-to-end; the most they could do would be to participate in the international process. But that wouldn't result in proprietary, patent-able technology, so they instead they put resources behind fringe science that they can control. This is reminiscent of the Lockheed-Martin CFR[7], which also chose an approach which was extremely unlikely to result in a scientific breakthrough but which did yield a few patents for Lockheed-Martin.
Frankly, behavior which is only one step removed from straight-up patent trolling justifies a certain degree of cynicism.
Just because they also look into other alternatives doesn't mean they are ignoring the more mainstream research. They are approaching it from every angle, because as pointed out, any discovery here can help in other places too.
> Frankly, behavior which is only one step removed from straight-up patent trolling
Google, for as big as it is and as many patents as it owns, has never once used its patent portfolio offensively, so assuming that they want to patent-troll is extremely unfair.
These two projects look great, and are exactly the kind of projects where I can see Google making an impact. I think it is extremely likely that we will struggle to build working reactors until we have an "AutoCAD for magnetic containment" which is able to accurately simulate real world plasma physics, after which I expect that a working design will be discovered within a few years, and then constructed within a decade or two. This follows from the simple observation that the build-test cycle for real world reactors is something like a decade, and for simulated reactors something like 1 day - a 36500X decrease in iteration time is a game changer. Such a program is very non-trivial and no one today knows how to write one.
And that's exactly the kind of stuff Google is trying to accelerate. One tangential but highly relevant example is their work on predicting molecular properties [0]. They were basically able to use neural networks to speed up the simulator by 300,000x, taking something that would take hours to less than a second. As you mention, this entirely changes the way you work and iterate. Here's a video of Jeff Dean talking more about it [1].
How can something like this even conceptually make sense? I'm sorry, but can someone explain to me how a neural network could approximate physically-deduced laws of nature (as had-coded into the classical simulator as I imagine) "without any distinguishable error"?
This is the biggest bs I have ever heard, since an molecular physics simulator that is beat by a neural network can only just be unoptimized as hell.
> Some reactors JT-60 have been energy positive[3] but are not yet self sustaining.
Nitpick: I don't believe the JT-60 has actually gone energy positive. They achieved a point where they would be power positive if they had been able to use a 1:1 ratio of D-T fuel instead of their D-D fuel but handling D-T fuel is very complicated and the JT-60 facility doesn't have those capabilities.
You could argue that this is exactly the type of research that should be funded by private industry: high-risk, high-reward, and unlikely to get government funding.
High-risk is something that private industry rarely to never does (except on vanity projects). All of the venture capital in Silicon Valley is really chasing medium-risk, not high-risk.
Beyond big tickets like fusion, I would direct you to things like initial drug investigations (almost all done in University labs with government grants), or rocket development (almost all done under government grants). In all of these cases the initial development of things that are unlikely to work out are done on public funds, and once it looks like it is just down to the details (still a big undertaking) then private funding steps in. The pattern here is almost always: socialize costs/risks, privatize gains.
1. Google, with all of the subject matter expertise they can avail themselves of, is completely wasting their money on this research that has no hope of success.
2. It'd be a bad thing for the world if we got cold fusion, but Google made some money off of it.
To me it sounded like olooney was arguing a third option:
3. Google will fail to achieve cold fusion but possibly get some patents which will make them money, but do little to advance overall scientific progress towards economic fusion.
Google can do what they like, of course. Perhaps they'll even make some money off it and create some useful device. I doubt Google has any greater access to SMEs - which in this case would be academic physicists - than ITER or CERN. I do believe that a dollar invested in tokamak research is more wisely spent than a dollar invested in cold fusion.
>Google, with all of the subject matter expertise they can avail themselves of, is completely wasting their money on this research that has no hope of success.
This isn't actually as silly as it sounds. There's a certain culture around self-absorbed futurism alive and kicking within these companies that produces weird and nonsensical projects.
One only has to look at Ray Kurzweil who turned from genuine inventor into some sort of nutrition supplement cyber guru and apparently still is employed by Google. There's also a lot of general silliness in the 'smart city' space, or cryogenics and lifespan enhancement. What all of those projects usually have in common is that they predominantly sound like something out of science fiction novels that the people would read who go on and found tech companies.
I am aware that cold fusion experiments are both much cheaper, and if successful, considerably safer and cheaper than hot fusion. Yet, there is no plausible mechanism for fusion to occur at room temperature. For two protons (hydrogen ions) to fuse, they must be brought to within 10e-15 m of one another. This requires overcoming the Coulomb repulsion between the two positively charged protons. This in turn requires a great deal of energy, roughly 6 KeV even once quantum tunneling is taken into effect. This corresponds to velocities of roughly 10e6 m/s, or roughly 0.5% the speed of light. This is is called the Gamow energy or the Gamow peak[2]. In hot fusion, this is accomplished by heating a gas to something like 10e7 K, at which point the average energy of any given proton in the plasma is roughly 6 KeV. Not coincidentally, this is also roughly the interior temperature of stars. It has been shown time and again that fusion does occur in a tokamak reactor... just not quite fast enough to overcome the cost of creating and containing the plasma.
At room temperature, the fraction of protons traveling at 0.5% of the speed of light is zero for all intents and purposes. Thus, spontaneous does not normally occur at room temperature, which we can all agree on. For cold fusion to proceed, there needs to be some other mechanism capable of making up this difference, either by somehow accelerating protons to an extremely high velocity, or otherwise encouraging them to fuse, perhaps by lowering the Coulomb barrier by some unknown mechanism. For the Fleischmann–Pons experiment (the original "cold fusion" experiment in the 1980s) this was hypothesized to be achieved by the crystal structure of palladium[3]. However, after the failure to replicate the original experiment, this hypotheses appears to have been falsified. In fact, no experiment has ever shown a measurable rate of fusion occurring at low temperatures.
And yet, there is the precedent of the Gamow factor. The classical potential for the Coloumb barrier is 3.4 MeV. Therefore, in a purely classical model, it is literally impossible to get two protons to fuse unless they collide at close to the speed of light. Yet Gamow showed that fusion could occur, with some probability, at much lower energies, thanks to a well known phenomenon called quantum tunneling. Why could there not be some other way taking this further? Furthermore, there is the precedent with fission. In the 1930's, many notable scientists were fission could ever be used as a power source. They were aware that fission could be effected by alpha particle bombardment, but this did not seem "energy positive." Sound familiar? And yet, when Hahn and Meitner[5] discovered spontaneous nuclear fission occuring due to a chain reaction in uranium in 1938, it immediately became apparent it could be used as a massive source of energy, and Einstein immediately warned FDR and the Manhattan project began.[4] Why could not a similar story play out for fusion? Cold fusion is tantalizingly plausible and the experiments, as you say, can be conducted on a tabletop. Why not try?
My answer, which is of course a subjective judgement, is that good science happens by searching where the light is, exploring the implications and edges of existing theories, not out in the dark, trying things completely at random. Rutherford wasn't bombarding gold foil for the hell of it. Randomly trying things in an atheoretic way in the hopes that a previously undiscovered and unsuspected piece of new physics will drop out is closer to alchemy, and just about as likely to be successful. More than 2,000 years of randomly combining urine and lead resulted in not one ounce of gold. (I grant you that tabletop cold fusion experiments may very well find a new bit of chemistry.) But the argument against is simply this: 6 KeV. It's simply too much. No chemical or electrical process is going to get you that, unless it first turns your experiment into plasma, in which case you're back to hot fusion! It's like throwing pebbles at the moon and calling it the Apollo program. It's not a matter of just getting the right pebble. You can try quartz and obsidian, rough and smooth, for as long as you like, but you're not even beginning to address the invariant in the room, which is that you just can't impart enough kinetic energy to your pebbles to even get them out of Earth's gravity well, much less to the moon. That is why I believe that a dollar spent on cold fusion actually has lower expected payoff than a dollar spent on hot fusion.
Coulomb forces are much weaker at the ends of highly elliptical nuclei. We've always assumed atomic nuclei are spherical. They are not. Iron nuclei have ends where the Coulomb forces are two orders of magnitude lesser than assuming a spherical shape would predict.
It would be a bad thing for the world if we got cold fusion and Google controlled the entire stack, yes.
This is because in the long, long term, allowing a company to dictate how a technology this important is used could hamper our ability to build on it or improve it. Over a long enough timescale, it would be better if we delayed getting cold fusion so that we could get the same technology later in a less encumbered form.
If cold fusion is possible, and Google stops working on it, we can still get it through other means. Other people can pick up Google's slack. If Google manages to crack it, and as a result is able to put legally enforceable restrictions on how it's used, then that's it. There's nowhere to go from there.
I have a feeling that if Google is successful at something along the lines of cold fusion, most governments will ignore any patents. The genie will be out of the bottle.
It's also assuming they would do something untoward with the monopoly, rather than just building a thousand new power plants and then making ten trillion dollars by being the lowest cost provider of zero carbon power generation for twenty years.
I don't know how the math works out exactly, but it seems not unlikely that the optimal rollout for curbing carbon emissions as rapidly as possible is not the same as the optimal rollout for maximizing value to shareholders.
I'm sure there's lots of ways it could happen. Off the top of my head - you can't just snap your fingers and start selling power, you need to build the reactors, scale out infrastructure, etc etc. There's huge expenditure involved. If you have a giant pile of money in the bank, earning interest, it might not be worth your while to spend it on that.
Another point - how much do you charge for the power? If your goal is to get people to switch, the answer is "as little as you can". If your goal is to make money, the answer is "as much as you can".
> If you have a giant pile of money in the bank, earning interest, it might not be worth your while to spend it on that.
But that's the case either way. If the technology was public domain, somebody would still have to pay to build the reactors, and then it's even harder to raise capital because with more competition there is less profit.
> If your goal is to make money, the answer is "as much as you can".
But "as much as you can" really means "just under what existing alternatives cost" -- because getting people to switch is how you make money. You don't make money by losing customers to competitors.
I'd believe Lockheed Martin's CFR claims more if they didn't say things like this:
> Energy created through fusion is 3-4 times more powerful than the energy released by fission.[0]
Which is just confus(ed/ing) on a fundamental level... More Powerful energy? So each Joule from fusion is actually 3-4 times better than each Joule from fission, in some way? I really can't work out what they are trying to say here.
But even that is pretty vague, since a single Uranium atom fission event releases about 200 MeV of energy, whereas most of the Hydrogen/Deuterium/Tritium/Helium/etc. fusion pathways release around 10% of this per fusion event: D + T -> He fusion gives 17 MeV, 4 p -> He gives 27 MeV, D + D -> He is 12 MeV and so on. Now, the energy released per nucleon greater for fusion, but I'm not convinced that's a useful or intuitive metric?
OK, so 235 kg of U_235 has the same number of fissile units (N atoms = 1000 x N_a) as 5 kg of D-T has of fusible units (2N atoms of combined weight 5 amu, or 1000 x N_a pairs) so (ignoring the N_a for now) one kg of U_235 gives 200 MeV / 235 = 0.85 MeV and the D-T fusion gives 17 MeV / 5 = 3.40 MeV
So, yeah, that's exactly four times the energy per kilogram. Now, why couldn't they just say THAT on the web page instead of the confusing and meaningless "this energy is 3-4 times more powerful than that energy" nonsense? But, thanks 'mikeash for clearing that up for me!
Tokamak[1] (and other magnetic confinement) reactors seem much more promising and have seen slow but non-zero improvements for decades[2]. Some reactors like JT-60 have come close to being energy positive[3] but are not yet self sustaining. Large ongoing projects like like ITER[4] have the potential to break past limits and provide invaluable data. There are also advances in plasma physics which improve our ability to run computer simulations[5] of Tokamaks and suggest new hypotheses for how they can be improved[6].
If Google were serious about fusion, they would invest in this kind of mainstream research. Unfortunately, that's too big for them to own end-to-end; the most they could do would be to participate in the international process. But that wouldn't result in proprietary, patent-able technology, so they instead they put resources behind fringe science that they can control. This is reminiscent of the Lockheed-Martin CFR[7], which also chose an approach which was extremely unlikely to result in a scientific breakthrough but which did yield a few patents for Lockheed-Martin.
Frankly, behavior which is only one step removed from straight-up patent trolling justifies a certain degree of cynicism.
[1]: https://en.wikipedia.org/wiki/Tokamak
[2]: https://en.wikipedia.org/wiki/Timeline_of_nuclear_fusion
[3]: https://en.wikipedia.org/wiki/JT-60
[4]: https://www.iter.org/
[5]: http://www.psfc.mit.edu/research/topics/plasma-fusion-theory...
[6]: https://phys.org/news/2019-01-scientists-stabilizes-fusion-p...
[7]: https://en.wikipedia.org/wiki/Lockheed_Martin_Compact_Fusion...