That's a version of "The Wait Calculation". For instance, it would be a waste of resources to build a starship right now to visit the closest other solar systems, because given our current maximum velocity and our rate of technological increase, it's a virtual certainty that in the future we would be able to build a second starship that would catch up and overtake it.

Eventually there is a break even point (for the closest solar systems, last I checked, it's around 400 years), but in order to know what it is you need to know the amount of time to reach the goal and your rate of increase. I don't know if there's a way to estimate that with fusion research.

This is a great mental model. I'm curious about applying this in areas of one's own life. Certain things that you'd actually accomplish faster, by starting later and leveraging the compounded time / resources that would have been spent pursuing it earlier.

The 80,000 hours folks generally apply this kind of reasoning to explain why it’s better for most people to just focus on a corporate job, stockpile a ton of cash, and donate it all when you die, instead of quitting to join a cause / non-profit - if your only goal is to optimize your impact.

Obviously people are free to choose differently and there are exceptions, unique circumstances etc.

I’m reasonably confident your summary of their advice is out of date. They now believe most organizations are talent constrained and the difference between the best and second best candidate is often very large so most people should be trying to work on hard problems fast if they can.

OT, but if solar panels last for over 20 years and get at least 5% cheaper every year, you save more money by waiting to install them than you do by installing them.

That does not factor in the money you make with the solar panel. I.e., if having a solar panel nets a profit of 10% of the cost of the solar panel, than installing one now makes more money than waiting a year.

ITER is not cheap. But it is expected to get to first plasma by 2025. There will be things learned at ITER that will be applied to SPARC and other smaller reactors. ITER is expected to cost a bit over twice what a fission plan costs. Considering the stakes, spending that money to learn things about fusion power generation seems well worth it.

I didn't see the details but it looks like both ITER and Sparc are both Tokamaks performing D-T fusion.

And in fact, it looks like Sparc is more of an exercise in making fusion reactors cheaper, most notably by using different magnets. ITER is more about validating the concepts, with no regard for the price. Besides the fusion itself, ITER will study how to inject fuel and evacuate the waste products as it is running. They will also tackle the problem of producing the large quantities of tritium a fusion plant requires.

So, my understanding is that Sparc is not intended to make ITER obsolete, but instead tackle a different problem. If commercial fusion reactors happen one day, I expect they will have a bit of both.

ITER is a research project and will likely be very valuable, even more so when the compact reactors turn out to be viable. As amazing as the compact reactors are they still face many of the same challenges as ITER.

ITER is a make-work project, with the goal of demonstrating a brave new world of international coöperation. Any research which emerges from the maelstrom is a nice bonus.

> But if you’d taken the ‘Manhattan Project’ approach and thrown 10x as much money at fusion research 20 years ago we’d most likely have spent most of it on a super-ITER.

Or maybe someone would have had enough sense to think "you know, what we really need is better superconducting magnets. Let's throw 1% of the budget at that problem."

Sure, but on the other hand, such a lucky break of this new approach might have had never happened. At the same time, if super-ITER approach was shown to work, by producing it en masse, it would likely be cheaper per installed MW than nuclear power, which would remove the need for solar panels and wind altogether. We'd have cheap power with no danger of nuclear pollution, no real nuclear waste problem, and no destroyed habitats and landscapes by solar panels and wind turbines. We'd still have won over current status quo, just not as much as we might with this lucky break.

Maybe, or maybe not, who knows? That’s the problem, your betting everything on your best guess what the optimum approach might be long before you have any idea what the odds are.

The problem is (1) the earlier you spend big the greater the chance you’ll pick wrong because you have less information to base a decision on and (2) Spending big doesn’t actually improve the odds that the approach you pick will end up being the right one. It just means you find out sooner.

You don't know that. What if you had spent 1000x the budget of the manhattan project on building an atom bomb in the year 1789? You would have gotten squat.

I don't think that's quite an apt comparison. The fundamental physics of an atomic bomb were not well known until the late 1800s. Apparently the Manhattan Project cost about $23 billion in 2007 dollars. ITER apparently had an initial budget of around $5 billion in 2005-6. By now, the construction cost is estimated to be at least $22 billion. It seems plausible that if we throw $15 or 20 billion at it in 2006, we could get bring forward the timeline by a handful of years. Or we might burn $20 billion on things like advanced construction and salaries for physicists and engineers, with relatively little to show for it.

The GDP of the EU is apparently $18 trillion, so $20 billion is about .1% of the GDP in a single year, and of course that $20 billion is spread over 1-2 decades. If anything, it seems like as a species, we should have more of these bets going. What if we spent 1% of our GDP on 10 long-shot, high-impact projects? Or hell, half a percent on 5 long-shot projects, and half a percent on solving the dozens of problems that we could solve simply by funding them.

Yes exactly, this is what Nassim Taleb writes about in Anti fragile. Bets with a known limited downside and the potential of an almost unlimited upside.

The upside is in no way unlimited. Even in the best case scenario, fusion power will be expensive. It might end up being useful for base load but it's unlikely to be cheaper than solar in most areas.

Assuming that you use it like before. But if you have a lot of power in one place you may think about using it for something like an orbital launch system.

The power grid already allows us to have as much power in one place as we want.

> The fundamental physics of an atomic bomb were not well known until the late 1800s.

Er, don't you mean the early 20th century?

Rutherford's exegesis of the nuclear model of the atom was published in 1911.[1]

1. https://en.wikipedia.org/wiki/Rutherford_model

Heh, wasn't sure where to draw the line. Radiation was discovered in the 1890s, so that seemed like as far back as you could go.

I know that, because we already got ITER, which is widely expected to be energy positive. Moreover, the ITER approach is known to scale upwards: in fact, the ITER is so large, because it's the smallest size that is expected to work.

ITER is expected to be energy positive, but it's a kind of pyrrhic breakeven. The gross power density of ITER will be 1/400th power density of a PWR reactor vessel.

Do you think that a better approach, given a lot of resources, would be to push more general/fundamental research without a clear end goal in mind? That approach makes sense to me intuitively given how many surprising breakthroughs there have been throughout history, but I admittedly know very little about actual research.

Most surprising breakthroughs (am not a historian though) were found with a clear goal in mind. It was just that the goal was very different from the breakthrough.

Ah ok, thanks. I guess in that case a good strategy could be to pursue a lot of specific goals, even if they don't seem worth it on the face of it. Seems like an approach the Ig Nobel awards would appreciate.

Probably not. REBCO was discovered in the 80s but is difficult to work with. A fusion energy Manhattan project would make 3-4 reactors in parallel. If it happened in the 2000s then they would have thrown significant effort at HTS technology. That’s one of the most obvious areas to invest in. The whole world would benefit from that.

ITER and Sparc are not mutually exclusive routes. ITER will teach us a lot about Fusion, knowledge that will be useful to scale up production.

"would now be just as redundant and superseded by this new approach."

And this approach will be superceded by something else. And that will be superceded in turn. Should we wait forever? We'd still live in caves.

These new approaches don't just show up, they come from the knowledge and scientists trained in whatever the previous approaches were.

Lastly, if this approach results in having cheaper fusion power, but needs extra 20 years of R&D, thats time we don't have.

It worked for WW2 and the nuclear bomb. It is hard but also effective.