Hacker News Reader

Comment by jlebrech | Hacker News Reader

nuclear fusion is always 10 years away.

I thought it was nuclear fusion was always 50 years away that sounds like a lot of progress.

PS: Considering it was 50 years away in 1980 we seem to be on track. Mostly though it just does not get much funding.

I watched a video about fusion where they say we have better superconductors now - the so called REBCO tape, that are cheap and most importantly - can generate stronger magnetic fields. So they can shrink the reactor 10x and make it cheaper and faster than the ITER project.

Source: https://www.youtube.com/watch?v=KkpqA8yG9T4

Sounds like they are ignoring a huge swath of problems.

So yes, with a lot of effort you could design a smaller and less expensive to construct structure assuming higher field strengths, but it would take massive amounts of time and R&D to get to that point.

IMO, we really just need to bite the bullet and build something at scale. Version 1 needs not be efficient, but it's going to teach us a lot more about operations than simulations can.

Reactors similar to MIT's ARC proposal have been built in about four years. You can do more experiments, do them faster, and since it's so much cheaper you can do more of them and take more risks. Speeding up progress is one of MIT's main objectives.

building something _at scale_ is basically the approach ITER has taken, and there is a different kind of problem in it: people & project management at that scale are very tough. Maybe tougher than the engineering even is.

When you need many countries to smoothly co-operate on a many-billion-dollar decades-long project that's hard to swiftly articulate or grasp, let alone pivot, while technology and geopolitics swiftly shifts around you, you're going to have a bad time.

ARC with modern superconductors is in many ways ITER but capable of being built on a multi-university budget, by virtue of those superconductors being much more powerful while being much smaller and lightweight and operating closer to room-temperature. The whole reactor needed for the same energy scale a ITER is physically much smaller.

The PSFC at MIT (who is proposing this plan) built the last great tokamak that got us closest to real fusion back in the 90s and that inspired ITER. ITER was the best plan available after that: the project size was necessary given conditions at the outset. Now many people are committed to it, and we're many billions in, so it likely isn't going away.

But ARC is probably a better version of basically the same plan given today's outlook: a scaled up tokamak, doable at university, national-lab, or skunkworks-level project sizes.

So I believe we could see huge progress from this in the next 5 years.

The only thing I have not seen articulated in Zach Hartwig or Dennis Whyte's otherwise fantastic talks in the past year is: what's next? What are the concrete steps between today and some version of ARC? They don't seem to be raising money yet, so it is clear there are some open questions about the other innovations in the ARC design, and I want to hear how/when we expect those to be solved or at least well-defined.

The problem in my mind is ITER is not really viable to produce power at reasonable costs. To be useful it would need to produce something like 5GW of heat for similar costs. Thus, ARC needs to be more than 10X as good to avoid hitting ITER scale in production.

the wright brother's wooden plane isn't viable at long distances.

Fusion progress chart:

http://imgur.com/BN0pz

Extrapolating this graph (whose last data point is approximately 1998), we should expect a commercial reactor very soon--perhaps in about 2005.

Looks like there were some setbacks, but these things happen.

I'm building one with '80s technology. That way, not only will I complete it, but I'll be able to include 20 years or so of improvements before I even turn it on.

There's an implicit claim there that the future will be like the past. Ie it has always been like this, thus it will continue to be so. What's your argument for that?

(I'm not making any claim about whether it will be it not)

...just like AGI

Reply on news.ycombinator.com