Re: Middle age bloat
Medieval warm period: my bingo card is filling out nicely...
29 publicly visible posts • joined 13 Jul 2021
Anyone who knows enough about massively parallel computing and networks operating over large distances with the unavoidable long latencies knows that this idea will not pan out very well. The scale-up in performance will be much less than the scale up in hardware required.
The problem would be less bad with a totally feed-forward algorithm (which the training of NNs is not), but you still have the feedback inherent in the network communication protocol.
Many of these AIs (LLMs in particular) are reaching a bottleneck already, having exhausted the available training data anyhow.
But they'll still manage to fool the investors for a while.
I think the point is that a person learning a subject or learning a "school" of artistry puts in work and gets some recompense for that work if they make products with it, or if they use their skills in employment for another person or company. Once you've trained an AI to do that, those people's ability to earn has been taken away, or at least significantly reduced. Companies get to have cheap results. Employees get to lose their jobs.
This has happened over the years with manual labour type things, via mechanisation and automation. It should be a good thing - freeing up people's time, allowing them to do more interesting and rewarding things. But instead it just drove real wages down, productivity up and corporate profits up. Now the same approach can be wielded against the more interesting and rewarding things. Sigh.
While I have some sympathy for the reasoning here, I think there's still a big issue with allowing even the public to benefit for free from tools which are in part trained upon copyrighted work. Buried in all that scraped personal data is work for which artists/creatives of various types are due royalties or at least control over how their hard work is exploited. These companies seem to have no regard for any kind of IP law. Better to delete the models and enforce real (not fake green-washing) carbon cost to the companies. Make them plant forests.
I agree that NIF is a terrible approach for power production. Your comments regarding Tritium breeding are off though. At least in theory. Test modules at ITER are planned for this, and a number of the private companies in this space will be getting to that stage in the next few years (probably before ITER). Beryllium or Lead can be used as a Neutron Multiplier (one in, two out kinda thing) and closed cycle gain of around 1.2 is predicted.
Surely you can put some numbers on this with back-of-the envelope calculations. You don't need to set up that experimental road, or even use a GPU. What is the per square km heat output of human activity on average? How does it compare with the per square km insolation from the Sun? Genuine question.
Oh, and the issues with high energy neutrons aren't just activation of reactor vessel materials. There's also blistering and embrittlement of materials over time. Besides the structural elements, Helion will have to replace their electromagnetic coils fairly frequently. They are very close to the reactor "core" and they are taking quite high forces. Their conductivity will start to suffer first and then, ultimately, their structural integrity too.
Regarding your other question about production of 3He, there are 3 sources.
Fusion of Deuterium with itself has 2 outcomes of equal probability: Tritium and a proton, or 3He and a neutron. So they can use their reactor design, if it really works, with DD fuel and then refine / separate the outputs. This is likely to be a net negative power reactor, so the secondary D-3He reactor that follows not only has to break even on its own, but also make up for that pre-processing loss.
They can sell the Tritium to other fusion companies, or sit on it and wait for it to slowly decay into 3He (12 year half-life). That's the second way to source 3He. They could procure Tritium (difficult and super expensive) and wait for it to decay. Tritium is currently produced in very small amounts in specially prepared fission reactors and is proposed to be self-generated / recycled in some future fusion reactors not of the Helion design. Both of these require 6Li and neutrons.
There's more 3He on the Moon than on Earth (or at least we think it will be more easily procured there). However the efficiency of mining the Moon for resources to use back on Earth is rather questionable, not to mention many, many decades away. This is the only source which does not require machinery to endure a high neutron flux. Just cosmic rays...
The trouble is scaling renewables and storage to cover everything including transport and industrial heat, not just domestic 'leccy just wont cut it without killing the planet with mining. There has to be a mix of carbon-free sources, some of which are immediately despatchable and can also be 24-7. Then there's the issue of bringing the third world and others up to the living standards us lucky few enjoy. Some kind of nuclear has to be in that mix. I don't see any alternative, besides perhaps space based solar, but that's another huge can of worms and isn't secure.
Unfortunately, that's not how two plasmoids rammed together and magnetically confined behave. Think of it more like a bunch of angry bees and a bunch of angry wasps in a small chamber. Bees will collide with bees as much as bees collide with wasps. They're whizzing all over the place, bouncing off each other and steered by magnetic fields. It is not a one-shot head-on collision at all.
With respect, have you researched Helion's process using sources of info other than Helion?
The 3He production from D-D is neutronic.
Further, D-D side reactions (neutronic) are unavoidable in any 3He-D bult plasma/plasmoid at 100 million Kelvin. Just look at the respective fusion cross-sections.
There is even the possibility of residual Tritium (you'll not evacuate it all between pulses), formed in the D-D side reactions then giving you some D-T fusion too. This means more neutrons, and higher energy ones at that.
The process is not aneutronic. D-D side reactions are unavoidable at the energies needed to get the D-3He reaction to work.
The 3He production process is also not aneutronic.
Problems.
Big problems with regulation and longevity.
However, unlike other approaches, the temperature issue is lessened. They are doing small scale repeated pulses. That should be easier to manage.
Oppenheimer and others had concerns that the nuclear *fission* bombs might cause a propagating chain *fusion* reaction (of hydrogen and nitrogen in the atmosphere and hydrogen in the oceans). Oppenheimer tasked Hans Bethe and others to calculate if this was actually possible. They concluded it was not. Indeed no A-bomb or H-bomb ever ignited the atmosphere.
If it was that easy, without confinement, we would have had fusion power stations decades ago.
The plasma in a tokamak is really not very dense. There are just a few grams of deuterium and tritium in there at any one time. Magnetic confinement is required to try to keep the plasma from melting or vaporising tiny holes in the first wall of the vacuum vessel, and to try to squeeze the plasma more tightly to increase the fusion rate. It is not necessary to prevent a chain reaction. Controlled fusion very rapidly fizzles if the conditions drift even slightly away from optimum.
You really don't need to worry about a Tokamak, or similar, turning into a nuclear bomb.
Do you have any reliable sources / references for this thermodynamics defying trick that you allude to regarding cavitation? There have been a lot of "woo" / fraudulent claims in that sphere over the years. Although there are (flawed) patents going back to the 70s claiming to be able to use it to enable fusion break even, I don't think I have seen anything convincing about cavitation in and of itself being an energy source, merely an energy concentrator or possible route to more efficient steam generation. Thanks in advance.
The sonoluminescence from the cavitation elicited by the pistol shrimp does indeed include the creation of a plasma, and fusion would not be absolutely out of the question, but the probability would be vanishingly low for even one fusion event, given the "triple product" we could infer from the conditions and the "fuel" species present.
Nevertheless, the pistol shrimp is the inspiration for First Light Fusion in Oxford. Their approach is all about focusing and amplifying shockwaves initiated by a physical projectile.
No disrespect, but if you knew what you were looking for then your google-fu needs a workout. Google search did used to be much more powerful and targetable than it is now, sadly, but you can still reduce the greediness of its matching by using quotation marks. Search for "Sail" "Stanford", for example.
In terms of fusion fuel, sure: less mass than you would need fission fuel for the same output. If you were fusing Deuterium and Tritium and had a closed loop Tritium cycle via a Lithium breeding blanket, you could stockpile enough Deuterium and Lithium and your Tritium seed load.
With a "solid first wall" reactor vessel you are probably going to have to do some maintenance well before 5 years though. There are some liquid first wall ideas out there like First Light and General Fusion which, if they can be made to work, would massively extend your maintenance intervals.
But I am not sure why the need for a sealed, set-and-forget fusion reactor. The materials and radiation levels involved are nowhere near as monster evil bogeyman as in fission reactors, and the nuclear proliferation issue is pretty much non-existent too.
We're straying further off the topic of the article now though.
The issue of SMRs creating 30 times the waste of a gigawatt scale power station is, perhaps, ameliorated by two future factors:
1) some gen 4 fission reactor proposals can be run from the waste from these SMRs and older plants. Overall the amount of net waste at the end goes down
2) fission may only be a relatively short-lived "bridge" now until the boffins get fusion to work properly and at commercial scale
Public and private investment is pouring into Fission both large scale and small modular on both sides of the Atlantic, but especially in the US. Similar can be said for fusion research.
I don't think we have a flotilla of energy Theranoses about to run aground out there. Investors are a little more scrutinising these days and while many startups may be chasing pipe-dreams, a few will eventually succeed.
The jokes about cow farts fall rather flat(ulent). Methane doesn't come out of the back end. It's a bovine burp thing, people. Plus, by itself, methane is odourless. Which is why they need the fancy IR satellite thingummy and not just a horde of roaming noses. Sorry to have caused such a stink.