I've used AI for two tasks where I found it very useful, and some very minor ones as well.

1) I had to write some code to invert a matrix in C. I knew the code was out there, but Google's search is so polluted today I could not find it. ChatGPT immediately returned working code. I noticed it did not calculate the determinant, so I asked it for that, and it modified the code to do so. As I say, I know that code is out there somewhere in a book, probably a dozen books, but I can no longer find older topics because the search engines are so polluted today. So yeah, this was extremely useful.

2) as part of the same project I wanted to OCR the 1975 book 101 BASIC Games. Normal OCRs are completely useless, they do not understand monospaced fonts or dot-matrix, and got perhaps 50% of the characters correct, or less. I ran out of tokens in ChatGPT, so I moved to Gemini. By telling Gemini that it was BASIC code, and in some cases quirks of the particular dialect, the accuracy was over 90%. It had problems with certain things - and other engines had *exactly* the same problems - but once I learned the problem areas I was able to track down the issues quickly. This saved me a LOT of time and I was able to OCR the entire book in a couple of nights.

3) with the first wave of tariffs there was a lot of news about the US's trade deficit. I found lots of hits in Google but again, hard to understand. So I asked ChatGPT and it made a very useful summary. My curiosity piqued, I then asked it for Canada's, and found to my surprise that it is actually much higher on a per-capita basis. But then the neat part - ChatGPT added a second paragraph noting that I can't simply compare it to the US's number, because Canada actually owns enormous foreign holdings, and that the resulting balance is largely flat, meaning the Canadian overall balance is pretty flat whereas the US is heavily indebted because their money normal turns inward.

There is simply no way I would have even known to ask that question. I learned a very important detail about all of this that I would not have otherwise.

So yeah, I think they can be extremely useful, if they are useful at all. My feeling, based on my limited use, is that the key to using them effectively is to ask them to perform very specific actions where the total context is limited. Write a whole program? I can't imagine it can understand the complexity. Comparing the debt to GDP ratio of several countries? No problem.

> Storing a kWh now just adds a few cents to its cost.

It's more than that, it roughly doubles the cost of the install.

Of course doubling the cost in China means it goes from about 80 cents to 1.60, which means it's still 1/4th the cost the same dispatch capacity as nuclear, which is why their nuclear program is 1/5th of their original plan.

Does anyone know the technical details here? I see "playable ads" in most games on iOS, and it sounds like this was similar. I assume these ads are also playable on Android, but maybe using some other tech?

I was under the impression that there are very many species of mosquito, and only some of the bite humans.

If the total number of ones that do is only a small percentage of the total, wouldn't an intermediate solution be to breed away (etc) the ones that bite humans? That would seem to have minimal side effects as compared to wiping them out universally.

> as the coins cost more to produce than their one-cent value

I'm sure this will fix the 2 trillion annual deficit. Now we can afford those tax cuts for rich people!

> You might want to show some evidence for that.

Which bit?

The bit that "solar/wind/nuclear/hydro can provide almost all the electricity needs"? Because that is self-evident given you are reading this.

Or do you mean the "fusion as a commercial power will never be a reality". That's less self-evident, especially if you have not worked in the industry, but here is a summary of the issue on that front:

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fmatter2energy.wordpress.com%2F2012%2F10%2F26%2Fwhy-fusion-will-never-happen%2F

> If General Fusion doesn't come through, a lot of money will be flushed down the toilet and a number of careers will go with it.

Sure, after over 20 years of being paid to do it. I don't think anyone will be crying on the way to the bank.

It is worth pointing out that this company was founded by a guy whose experience in this field was working on ink jet printers.

You think I'm joking, right?

> Lockheed Martin somehow got away with quietly hiding their reactor

LM's reactor design was subjected to peer review which demonstrated it would not work. They apparently built one model and it was much worse than even the predictions. Everyone working on it was moved to other projects just before COVID. There is no "quiet hiding", everyone in the field that was following the project was aware of its "progress" all along.

> And then there's Commonwealth Fusion. And Focus Fusion. And possibly others!

No no no no no, a million times no.

There is an enourmous difference between CFS and the rest - well Tokamak Energy in the UK is similar to CF, but they're in the UK so no one talks about them.

You see, CFS and TE, are projects based on designs that have 50 years of experimental evidence behind them. Their performance is understood. Both projects differ from earlier efforts largely on the engineering side, they are both trying to come up with better ways to *build* the reactor. CFS is all about the magnets, and TE is all about the size. But the plasma inside is essentially identical to the one in JET and ITER.

In contrast, something like Focus Fusion has zero physics of interest. "Of interest" means "experimental physics demonstrating performance in the region needed to be an operational reactor". The same is true for Helion, TAE, General Fusion, Zap, and all the rest. None of them have the experimental evidence that suggests their design will work. And if one considers the 85 year history of the field, history says they will all discover new problems as they try to scale towards production.

To quote Jassby:

There are also numerous wannabe fusion scams that have popped up in the last 5
years or so, making the usual preposterous claims on the basis of nothing but hot air or
cold plasma, but these outfits are not yet sufficiently well-known to warrant more than a
mention. Examples are Dynomak, First Light, HyperJet, and numerous members of the
delusional Fusion Industry Association.

The long and short of it is that these companies are built on hot air of precisely the wrong type. But in the USA, the idea that private industry is better at everything than the government is a matter of assumed truth, and therefore simply saying "we're private, thus better" will result in everyone believing you, and in some cases, giving you money. Lots and lots of money.

Let me give you one example.

A couple of years ago I heard rumours of a paper that demonstrated why TAE's approach will never work. For background, TAE formed in 1998, claiming breakeven in three years and commercial demo in five. They have been making this claim continually for the last 27 years. What they don't mention is that just before they went public, they approached the Naval Research Laboratory (which does fusion research) for funding. I contacted one of the people and he sent me the paper, which states in no uncertain terms that there is absolutely no way their concept can work.

But more than that, he described how the paper came to be. As part of the application process, someone in the lab will be assigned to do peer review. At this time, the two reviewers were new to the lab, so they got it dumped in their lap. They started working on it in the morning, and before lunch had already found (IIRC) eight absolutely killer problems. So they went to lunch. One of the two said the concept was so bad that he seriously wondered if the author was suffering from mental illness (by which I assume he means dementia).

But that's not all. I was contacting him 20 years after they wrote the report, a period in which TAE received several very large rounds of funding in the hundred of millions of dollars. The author noted that in those 20 years, I was the first person he can recall asking about their report. He went on to say that as far as he was aware, none of the people involved in the funding rounds had ever tried to contact either of them.

I believe this story is useful in explaining the current state of the VC market in fusion circles.

If you wish to read the paper, it is here:

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fapps.dtic.mil%2Fsti%2Ftr%2Fpdf%2FADA356110.pdf

If you google "rostoker colliding beam" you'll find many similar critiques in other sources as well, including Science and the IEEE.

> Helion is claiming to have reactors online by 2028.

Sure, and we should trust that prediction because:

In 2014, they said they would have commercial plants in 2019.
In 2015, they said breakeven in two years and a 50 MW commercial demo in 2019.
In 2018, they said breakeven in less than three years.
In 2023, they said net electricity export in 2024.

But I'm sure they are correct this time, because [INSERT TECHNOBABBLE HERE].

> Seems like the NIF is trailing private enterprise.

In bogus claims? Yes.

In actual fusion reactions? No.

> Some designs, like General Fussion's

Ok, but GF's design will almost certainly not work.

GF claims their idea is taken from an earlier project called Linus. They say, and you can find quotes to this effect all over, that the problem with Linux was that they didn't have the control systems needed to produce symmetric collapse. But today, with ASICs and such, they can do it.

So I tracked down some of the people that worked on Linus. They sent me photographs of their fully symmetric collapse. They said it was not even that difficult. They also explained that this was not what killed the project, but the introduction of the field-reversed configuration as a way to keep the warm plasma inside the machine. When they ran the numbers, it appeared that you didn't need the machine at all, because the FRC itself could do it all. So they all stopped working on Linus and moved to FRC.

GF has been around since 2003, IIRC. In those 22 years since, they have not run a single experiment that matches their proposed machine. They have built a few compression machines, only to change the design, and they have run a few maglif experiments using entirely different compression concepts, but not one single experiment that uses their proposed compression system with their proposed plasma system. In fact, those maglif experiments also do not use their proposed ST plasma layout.

Yet this is apparently the fast path to commercial fusion, which will give breakeven in three years and commercial prototype in five, something they have been saying since they formed.

> And yet no vehicles anywhere are propelled by bullets.

Rockets.

> No person has a greater role or responsibility for the creation of NIF that Tarter

Nuckolls clearly has a greater role and responsibility for the creation of NIF than Tarter.

Of course, you wouldn't know that if you read a book on the topic *by Tarter*.

> The NIF exists for nuclear weapons research

Before the NIF lit it up, LLNL was only too happy to tell everyone it was a system for commercial power production, something they have been saying about the ICF program since it was revealed publicly in 1972. They sold it as a lower-cost faster approach to commercial power than the magnetic approach, especially after their mirror program was killed in 82.

In fact, LLNL spent some money (ie, seven or eight figures) on a program called LIFE that was dedicated to converting the NIF-type layout to commercial production. It ran for five years and shut down only when NIF did, in fact, not work. You can read about it here:

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fen.wikipedia.org%2Fwiki%2FLaser_Inertial_Fusion_Energy

> In the 70s there were no lasers that could power this reaction

Well they certainly thought they did at the time. Shiva was expected to get close to breakeven, and Nova was expected to easily beat it. The calculations suggested only about 10 to 100 kJ would do it.

The calculations were wrong, of course, but they didn't know that at the time.

> Methinks you misunderestimate the potential for progress here

Well I've been following ICF since I was introduced to it by National Geographic in 1978. And given everything I know about it, I don't think anyone is underestimating it at all.

Let's run a couple of numbers here. Studies by LLNL during the LIFE project suggested that the fuel capsules might read 50 cents when in mass production. In NIF, they could produce a maximum of 50 MJ of fusion output. We would get 35% of that as electricity after conversion. So we might get 15 MJ output, but need 300 to 400 MJ input.

Now it is possible to improve the lasers so they are not so bad. In fact, we are pretty sure we can get 10x efficiency without too much trouble, and in theory that might get as high as 20%. So let's go with 20%. That means that we need 20 MJ to power them, which is still more than the amount of power we could hope to extract from the reaction.

But let's go all-in, let's say the lasers require no energy at all. So we get all 15 MJ out as electricity. Now you might not be very familiar with MJ, so let's convert to an everyday unit. 15 MJ is the same as 4 kWh.

I don't know what power costs where you are, but right now here in Ontario utility backbone power is selling for (hold on a sec) 4.11 cents. You can verify that here:

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fieso.ca%2FSector-Participants%2FMarket-Operations%2FLegacy-Market%2FHourly-Ontario-Energy-Price

So, 4 kWh at 4 cents each is 16 cents worth of power.

From a system that costs, absolute minimum, 50 cents.

So no, I don't think anyone is underestimating this approach.

The funny thing is that people have been saying this from the very beginning in the 1960s. It even has a name, the "kopeks problem". People in the labs that dared to talk about this were sidelined, so people stopped talking about it.

And so, a full SIXTY YEARS after they built the first machine, 4pi, we are just as far from commercial power from ICF as we ever were, and *absolutely everyone that works in the program is perfectly aware of that fact*

> since the Molten Salt Reactor Experiment but Nixon tried to bury it to promote
> conventional reactor building in his home state of California

Oh balony. He was all-in and publicly stated his support for the breeder program through his entire term. Here is his statement to congress on the issue in 1973:

Most nuclear power plants now in operation utilize light water reactors. In the near future, some will use high temperature gas-cooled reactors. These techniques will be supplemented during the next decade by the fast breeder reactor, which will bring about a 30-fold increase in the efficiency with which we utilize our domestic uranium resources. At present, development of the liquid metal fast breeder reactor is our highest priority target for nuclear research and development.

You can read it here:

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.presidency.ucsb.edu%2Fdocuments%2Fspecial-message-the-congress-energy-resources

Moreover, his administration, and those that followed, were such advocates of the breeder program that they were dead set against fusion as a result. I cannot find the quote now, but it's definitely in Heppenheimer, that they considered the whole fusion program "pie in the sky" and that it was a distraction from the breeder program and they believed it might cause some to shift support away from the breeders to fusion.

This was certainly still the case during the PLT weekend in 77 whenDean and Kintner nearly got fired for releasing PLT's results.

Heppenheimer's Man Made Sun goes into this in some detail, and is an excellent read on the topic of both fusion and the breeder program.

> I mean, just about every commercial jet is a lot bigger than a car.

Which is precisely why the energy requirements are tractable.

There are very few 4-person jets in the world, and those that exist are extremely expensive to operate. They scale with passenger capacity, which is why commercial jet designs continually scale upwards over time - consider the history of the 737 for instance.

... and since then, it has been demonstrated that the polywell basically didn't work at all. It might in theory, but when they built them the virtual electrodes the concept is based on turned out to be illusions, and when they considered that, the current needed to maintain them turned out to be infinity minus one.