6 ms·

Meta rant: I would like people to remind themselves that HN's rules explicitly disallow shallow dismissals. Unfortunately this comment section seems to be nothi

Meta rant: I would like people to remind themselves that HN's rules explicitly disallow shallow dismissals. Unfortunately this comment section seems to be nothing but. The material science behind high temperature semiconductors has undergone radical changes in recent years. The techniques being used are largely well understood science and the downsides of previous approaches like ITER are explicitly avoided. There has been a visible explosion of commercial activity in recent years. People bander around the "always 20 years away" moniker as if we should put today's estimates on the same grounds as the guesses from those early 1970s projects that were never even properly funded. People were not screaming "20 years more" at the turn of the century. You don't get to discount something as potentially groundbreaking as this just because it's taken a little while to get here. Either put up or shut up. I would love to hear genuine substantive conversations here, if people were willing to have them.

annywhey 9y ago
I can offer some economic perspective. Oftentimes the critical advance for technology to reach deployment is not a theoretical breakthrough in the primary science but rather a change in aggregate costs, ability to organize capital, secondary technologies, etc. It's stuff that pushes the tech past a threshold of feasibility. The Industrial Revolution itself, for example, was a mix of recent breakthroughs and older ones that started to be realized enmasse(e.g. coal mining and steam engines co-evolved during the 1700's). So when researchers say, "this supporting piece of the puzzle has gotten much easier," as it is with the new magnets described in the article, that's actually a good time to listen since it might be the tipping point. Theory developments or experimental observations are exciting stories to read about but often have less imminent bearing on whether you're going to see it impact your life.
- alehul 9y ago
  Could you recommend any resources (articles, books, podcasts, etc.) that go into more detail on the economic factors that come together to enable periods such as the Industrial Revolution, or really any period? I think economics offers a great perspective to subscribe to, but haven't found much opportunity to learn it. Most "economics" podcasts are really just business-focused and describe modern events with little insight into the economics behind it.
  - brownbat 9y ago
    The best fast and fun one is "How We Got Here" by Kessler, I think it's free if you send an email. Draws a really clever line of related inventions from swords to microchips. Specific areas after that... "Brilliant" by Brox on the history of lighting. "Salt" by Kurlansky. "The Master Switch" by Wu on telecoms and broadcasting. "Ascent of Money" by Ferguson has some critics, but it's got some great historical anecdotes on the history of financial tools, like fractional reserves and insurance, weirdly captivating. Those are all economic history... If you really want a pop intro to econ principles, maybe Planet Money? Pretty much everyone should listen to Planet Money anyway, it's just great radio.
    - anadem 9y ago
      Here's Kessler's excellent How We Got Here (pdf): http://www.andykessler.com/andy_kessler/excerpts/How_We_Got_Here_Andy_Kessler.pdf http://www.andykessler.com/andy_kessler/excerpts/How_We_Got_...
    - dmix 9y ago
      I gave Planet Money a shot but found the two hosts nearly unbearable and the typical NPR heavy-handed production and jumpy editing distracting. They try so desperately to make the topics sound fun/accessible that it ends up sounding like a disingenuous forced comedic act rather than a natural, education conversation with the guests...I guess I'm not the target audience for this. But I love the book recommendations. I plan to read that Kessler book tonight.
  - TheCowboy 9y ago
    I thought Douglass North's textbook on economic history was useful when I studied it in school, and he likely has a lot of other articles and books you can pick and choose from. He died in 2015 but was quite prolific and won the Nobel prize. I have the last book he co-authored on my shelf that I still need to read. I think he's most known for his work in the role of institutions (which I think helps elucidate the "economic factors that come together" part of your curiosity): https://en.wikipedia.org/wiki/Douglass_North#Institutions https://en.wikipedia.org/wiki/Douglass_North#Institutions https://en.wikipedia.org/wiki/New_institutional_economics https://en.wikipedia.org/wiki/New_institutional_economics
cctt23 9y ago
Unsolved problems for fusion as powerplant: -Sputtering of shielding and every other part of the reactor from fast neutrons. If your reactor is becoming brittle as it’s in operation, and requires constant maintenance, it won’t be operating enough to be cost effective. -Breeding blankets. If we’re not breeding tritium in the blanket (and no one has been able to sustain a reaction that way yet) then we’re just using fission a lot. Expense becomes an issue, as does radiological issues. -The plasma diverter is very much an unsolved problem. I can get into more detail here, but in short this is the part of the reactor that “skims” some of the hot plasma off to do the work. The dynamics of very hot, magnetically constrained plasmas still escapes us, and when you throw a rock into that stream, the complexity increases. Current divertes wouldn’t last a day in an operating plant. Disassembling your whole plant every day and reassembling it is a non-starter. -Containment of plasma at sufficient energies is still something measured in seconds, or fractions of seconds. The usual metaphor is trying to uniformly squeeze a balloon; it will just “squirt” out. For s research reactor a second or two of fusion is an achievement. For power generation it’s nothing. -Neutron activation of otherwise inert materials means you’re going to have serious radioactive waste. It’s unclear just how dirty D-T fusion would be from soup to nuts, but “pretty dirty” seems like a good bet. -Tritium penetration. -Most of the energy produced is in the form of neutrons, and we don’t know how to use that as a source of power. Those neutrons, in addition to destroying the reactor itself and activating materials, represent a loss. -What we really need is aneutronic fusion through alternative cycles to D-T, like p-p, but that’s a much hotter plasma and no one has a clue how to make it work yet. -Coolant for a constantly running reactor is a boring, but unsolved problem. There’s more, but these are the ones most poeple on HN probably are aware of when they dismiss this article. Some further reading https://thebulletin.org/fusion-reactors-not-what-they’re-cracked-be10699 https://thebulletin.org/fusion-reactors-not-what-they’re-cra...
- ISL 9y ago
  The waste problem from neutron activation is beginning to get the publicity it deserves in both articles and in comments like the parent's. Fusion is advertised as a clean power-generation technology, but it will, in general, have its own radioactive-waste disposal problems.
gerdesj 9y ago
I've had a quick shufti through the comments so far and I don't see too many "20 years" and anyway the meme was always 50 years. That said, I am with you - I think the 20, 25, 50 memes are badly wrong. "People were not screaming "20 years more" at the turn of the century." - which century? I went to school in Oxfordshire, UK (later half of the 1980s) and studied physics to A level and went on a field trip or two to the local nuclear fusion project. I clearly remember being told by everyone there how progress was measured. It seemed to boil down to a chart with time on the x axis and factors away from sustainable fusion on y. This may have been a "lies to children". The chart looked like an exponential decay with a rather wobbly tail starting in the late or so 1970s back then. There is a good reason for the 20/25/50 years meme. It's the best model we have had for progress for a very long time. I will be ecstatic to get my 'leccy from nuclear fusion in my life time and unimpressed if my grand daughter is still waiting.
MR4D 9y ago
May I quote from the end of the article: “Prof Wilson was also cautious about the timeframe, saying that while the project was exciting he couldn’t see how it would achieve its goal of putting energy on the grid within 15 years.” That’s deep enough for me.
- saudioger 9y ago
  > he couldn’t see how it would achieve its goal of putting energy on the grid within 15 years.” but in 16 years tho... maybe
brudgers 9y ago
My father was 14 when Ivy Mike blasted Elugelab below sea level. In 1982, I took my first college course. We read Herman Kahn's The Next 200 Years. Tokamaks were promising. I shared my seventeen year old enthusiasm for the fusion future with him. He introduced me to "Always twenty years in the future." He'll be 80 next month. The article's optimistic 15 years is 2033. For Dad, that would put it just over four "Twenty years in the future." For me, it would only be two-and-a-half. Sure there's a reasonable chance that the "always" in "always twenty years in the future" will be wrong some day. I'm skeptical this is that day because the basis for today's prediction is that the future is just like the present only more so. The article is a press release not science. The just-over-the-horizon timeframe normally cited is 30 years, but the MIT team believe they can halve this by using new superconducting materials to produce ultra-powerful magnets, one of the main components of a fusion reactor. Halving the just-over-the-horizon number isn't progress. It's marketing.
sunstone 9y ago
High temperature semiconductors and high temperature superconductors are quite different things.
godelski 9y ago
>People bander around the "always 20 years away" moniker as if we should put today's estimates on the same grounds as the guesses from those early 1970s projects that were never even properly funded. People were not screaming "20 years more" at the turn of the century. Well I think people have great reason to be skeptical. I hear "Fusion is around the corner" at least once a year. Usually it is by smaller companies but every once in awhile it is the bigger ones. So let's look at the big one. In late 2014 LOCKHEED MARTIN announced that their compact fusion device would be demonstrated in a few years and have an operational prototype by 2019 [1]. Even Scott Manley has the top comment on that video as ""high risk" = we don't actually have it working, but there's a chance we could." Even in the video suggests it wouldn't get onto the grid by 2034 (20 years from their announcement). So maybe you're right, it isn't 20 years away, only 16. Which serendipitously lines up with the 15 years claimed in the FIRST SENTENCE. Which is then given many asterisks at the end and doubted by the lead scientist. TLDR: People band around "it is 20 years away" because experts are still saying "it is 20 years away." (including this article) [1] https://www.youtube.com/watch?v=UlYClniDFkM https://www.youtube.com/watch?v=UlYClniDFkM End note: That being said, I don't want to downplay the improvements. We are SIGNIFICANTLY closer than we were in the 80's/90's. Several of the major problems have been improved upon. And I do think I will see fusion on the grid in my lifetime. More so, I really think it will happen within 20-30 years.
ggm 9y ago
As a serial offender, I think you're right to say this but there is the other side of the coin. Fusion is littered with giant claims, and huge funding holes. Skunk works are very cool but it's as far from contestable science as it's possible to be: the skunkworks doesn't want to be put underneath a critical microscope because its looking for exploitable trade secrets and patent material. Groundbreaking is usually overused. How many genuinely groundbreaking things emerge from press releases as opposed to deep investment by state backed actors? Seriously, Edison (for instance) was a wizard at commercial applications but frankly useless as a scientist. I call Leo Szilard groundbreaking. He really did have a flash of insight. Crispr didn't just happen. It took a shedload of prior work. FMRI is groundbreaking except they're all walking back from the pop sci stories into more fundamental tests. So is this stuff really worth substantial conversation? Who is going to invest in chatter here? MIT and Lockheed won't, for obvious reasons. And I doubt jet or iter scientists will, for other obvious reasons. Who is left who would meet your criteria?
Ice_cream_suit 9y ago
Semiconductors and superconductors are very different beasts.
appleflaxen 9y ago
I agree with you regarding shallow dismissals, provided the claims are deep to start with. In this case, we start with a headline that the most transformative technology since combustion is just around the corner. The basis for that claim? That they are going to create an experimental reactor using new high-temp superconductors. Well ok, great. But this is just a lame press release about how somebody HOPES it will work. It's not even a technical argument: it's basically a business argument, using currently available technology. Just the nuclear power version of the "suits" article that comes around every year, and that PG called out in 2005 [1] So what, exactly, can a person substantively say against /this/ piece? I would argue: practically nothing. The best points raised by others relate to fusion in general, which we could have had without this submarine PR piece. 1. http://www.paulgraham.com/submarine.html http://www.paulgraham.com/submarine.html
tutanchamun 9y ago
> The techniques being used are largely well understood science and the downsides of previous approaches like ITER are explicitly avoided. I'm a layman when it comes to these things, but does that mean they have improved tokamak's or will future nuclear fusion reactors more look like the stellarator used in Wendelstein 7-X for example?
dekhn 9y ago
Actually, you can easilly discount this, since they haven't run the experiment yet. Until that happens it should be fully discounted. Happy to pay attention when the experiment is run.

annywhey 9y ago

I can offer some economic perspective. Oftentimes the critical advance for technology to reach deployment is not a theoretical breakthrough in the primary science but rather a change in aggregate costs, ability to organize capital, secondary technologies, etc. It's stuff that pushes the tech past a threshold of feasibility. The Industrial Revolution itself, for example, was a mix of recent breakthroughs and older ones that started to be realized enmasse(e.g. coal mining and steam engines co-evolved during the 1700's). So when researchers say, "this supporting piece of the puzzle has gotten much easier," as it is with the new magnets described in the article, that's actually a good time to listen since it might be the tipping point. Theory developments or experimental observations are exciting stories to read about but often have less imminent bearing on whether you're going to see it impact your life.

alehul 9y ago
Could you recommend any resources (articles, books, podcasts, etc.) that go into more detail on the economic factors that come together to enable periods such as the Industrial Revolution, or really any period? I think economics offers a great perspective to subscribe to, but haven't found much opportunity to learn it. Most "economics" podcasts are really just business-focused and describe modern events with little insight into the economics behind it.
- brownbat 9y ago
  The best fast and fun one is "How We Got Here" by Kessler, I think it's free if you send an email. Draws a really clever line of related inventions from swords to microchips. Specific areas after that... "Brilliant" by Brox on the history of lighting. "Salt" by Kurlansky. "The Master Switch" by Wu on telecoms and broadcasting. "Ascent of Money" by Ferguson has some critics, but it's got some great historical anecdotes on the history of financial tools, like fractional reserves and insurance, weirdly captivating. Those are all economic history... If you really want a pop intro to econ principles, maybe Planet Money? Pretty much everyone should listen to Planet Money anyway, it's just great radio.
  - anadem 9y ago
    Here's Kessler's excellent How We Got Here (pdf): http://www.andykessler.com/andy_kessler/excerpts/How_We_Got_Here_Andy_Kessler.pdf http://www.andykessler.com/andy_kessler/excerpts/How_We_Got_...
  - dmix 9y ago
    I gave Planet Money a shot but found the two hosts nearly unbearable and the typical NPR heavy-handed production and jumpy editing distracting. They try so desperately to make the topics sound fun/accessible that it ends up sounding like a disingenuous forced comedic act rather than a natural, education conversation with the guests...I guess I'm not the target audience for this. But I love the book recommendations. I plan to read that Kessler book tonight.
- TheCowboy 9y ago
  I thought Douglass North's textbook on economic history was useful when I studied it in school, and he likely has a lot of other articles and books you can pick and choose from. He died in 2015 but was quite prolific and won the Nobel prize. I have the last book he co-authored on my shelf that I still need to read. I think he's most known for his work in the role of institutions (which I think helps elucidate the "economic factors that come together" part of your curiosity): https://en.wikipedia.org/wiki/Douglass_North#Institutions https://en.wikipedia.org/wiki/Douglass_North#Institutions https://en.wikipedia.org/wiki/New_institutional_economics https://en.wikipedia.org/wiki/New_institutional_economics

cctt23 9y ago

Unsolved problems for fusion as powerplant: -Sputtering of shielding and every other part of the reactor from fast neutrons. If your reactor is becoming brittle as it’s in operation, and requires constant maintenance, it won’t be operating enough to be cost effective. -Breeding blankets. If we’re not breeding tritium in the blanket (and no one has been able to sustain a reaction that way yet) then we’re just using fission a lot. Expense becomes an issue, as does radiological issues. -The plasma diverter is very much an unsolved problem. I can get into more detail here, but in short this is the part of the reactor that “skims” some of the hot plasma off to do the work. The dynamics of very hot, magnetically constrained plasmas still escapes us, and when you throw a rock into that stream, the complexity increases. Current divertes wouldn’t last a day in an operating plant. Disassembling your whole plant every day and reassembling it is a non-starter. -Containment of plasma at sufficient energies is still something measured in seconds, or fractions of seconds. The usual metaphor is trying to uniformly squeeze a balloon; it will just “squirt” out. For s research reactor a second or two of fusion is an achievement. For power generation it’s nothing. -Neutron activation of otherwise inert materials means you’re going to have serious radioactive waste. It’s unclear just how dirty D-T fusion would be from soup to nuts, but “pretty dirty” seems like a good bet. -Tritium penetration. -Most of the energy produced is in the form of neutrons, and we don’t know how to use that as a source of power. Those neutrons, in addition to destroying the reactor itself and activating materials, represent a loss. -What we really need is aneutronic fusion through alternative cycles to D-T, like p-p, but that’s a much hotter plasma and no one has a clue how to make it work yet. -Coolant for a constantly running reactor is a boring, but unsolved problem. There’s more, but these are the ones most poeple on HN probably are aware of when they dismiss this article. Some further reading https://thebulletin.org/fusion-reactors-not-what-they’re-cracked-be10699 https://thebulletin.org/fusion-reactors-not-what-they’re-cra...

ISL 9y ago
The waste problem from neutron activation is beginning to get the publicity it deserves in both articles and in comments like the parent's. Fusion is advertised as a clean power-generation technology, but it will, in general, have its own radioactive-waste disposal problems.

gerdesj 9y ago

I've had a quick shufti through the comments so far and I don't see too many "20 years" and anyway the meme was always 50 years. That said, I am with you - I think the 20, 25, 50 memes are badly wrong. "People were not screaming "20 years more" at the turn of the century." - which century? I went to school in Oxfordshire, UK (later half of the 1980s) and studied physics to A level and went on a field trip or two to the local nuclear fusion project. I clearly remember being told by everyone there how progress was measured. It seemed to boil down to a chart with time on the x axis and factors away from sustainable fusion on y. This may have been a "lies to children". The chart looked like an exponential decay with a rather wobbly tail starting in the late or so 1970s back then. There is a good reason for the 20/25/50 years meme. It's the best model we have had for progress for a very long time. I will be ecstatic to get my 'leccy from nuclear fusion in my life time and unimpressed if my grand daughter is still waiting.

MR4D 9y ago

May I quote from the end of the article: “Prof Wilson was also cautious about the timeframe, saying that while the project was exciting he couldn’t see how it would achieve its goal of putting energy on the grid within 15 years.” That’s deep enough for me.

saudioger 9y ago
> he couldn’t see how it would achieve its goal of putting energy on the grid within 15 years.” but in 16 years tho... maybe

brudgers 9y ago

My father was 14 when Ivy Mike blasted Elugelab below sea level. In 1982, I took my first college course. We read Herman Kahn's The Next 200 Years. Tokamaks were promising. I shared my seventeen year old enthusiasm for the fusion future with him. He introduced me to "Always twenty years in the future." He'll be 80 next month. The article's optimistic 15 years is 2033. For Dad, that would put it just over four "Twenty years in the future." For me, it would only be two-and-a-half. Sure there's a reasonable chance that the "always" in "always twenty years in the future" will be wrong some day. I'm skeptical this is that day because the basis for today's prediction is that the future is just like the present only more so. The article is a press release not science. The just-over-the-horizon timeframe normally cited is 30 years, but the MIT team believe they can halve this by using new superconducting materials to produce ultra-powerful magnets, one of the main components of a fusion reactor. Halving the just-over-the-horizon number isn't progress. It's marketing.

sunstone 9y ago

High temperature semiconductors and high temperature superconductors are quite different things.

godelski 9y ago

>People bander around the "always 20 years away" moniker as if we should put today's estimates on the same grounds as the guesses from those early 1970s projects that were never even properly funded. People were not screaming "20 years more" at the turn of the century. Well I think people have great reason to be skeptical. I hear "Fusion is around the corner" at least once a year. Usually it is by smaller companies but every once in awhile it is the bigger ones. So let's look at the big one. In late 2014 LOCKHEED MARTIN announced that their compact fusion device would be demonstrated in a few years and have an operational prototype by 2019 [1]. Even Scott Manley has the top comment on that video as ""high risk" = we don't actually have it working, but there's a chance we could." Even in the video suggests it wouldn't get onto the grid by 2034 (20 years from their announcement). So maybe you're right, it isn't 20 years away, only 16. Which serendipitously lines up with the 15 years claimed in the FIRST SENTENCE. Which is then given many asterisks at the end and doubted by the lead scientist. TLDR: People band around "it is 20 years away" because experts are still saying "it is 20 years away." (including this article) [1] https://www.youtube.com/watch?v=UlYClniDFkM https://www.youtube.com/watch?v=UlYClniDFkM End note: That being said, I don't want to downplay the improvements. We are SIGNIFICANTLY closer than we were in the 80's/90's. Several of the major problems have been improved upon. And I do think I will see fusion on the grid in my lifetime. More so, I really think it will happen within 20-30 years.

ggm 9y ago

As a serial offender, I think you're right to say this but there is the other side of the coin. Fusion is littered with giant claims, and huge funding holes. Skunk works are very cool but it's as far from contestable science as it's possible to be: the skunkworks doesn't want to be put underneath a critical microscope because its looking for exploitable trade secrets and patent material. Groundbreaking is usually overused. How many genuinely groundbreaking things emerge from press releases as opposed to deep investment by state backed actors? Seriously, Edison (for instance) was a wizard at commercial applications but frankly useless as a scientist. I call Leo Szilard groundbreaking. He really did have a flash of insight. Crispr didn't just happen. It took a shedload of prior work. FMRI is groundbreaking except they're all walking back from the pop sci stories into more fundamental tests. So is this stuff really worth substantial conversation? Who is going to invest in chatter here? MIT and Lockheed won't, for obvious reasons. And I doubt jet or iter scientists will, for other obvious reasons. Who is left who would meet your criteria?

Ice_cream_suit 9y ago

Semiconductors and superconductors are very different beasts.

appleflaxen 9y ago

I agree with you regarding shallow dismissals, provided the claims are deep to start with. In this case, we start with a headline that the most transformative technology since combustion is just around the corner. The basis for that claim? That they are going to create an experimental reactor using new high-temp superconductors. Well ok, great. But this is just a lame press release about how somebody HOPES it will work. It's not even a technical argument: it's basically a business argument, using currently available technology. Just the nuclear power version of the "suits" article that comes around every year, and that PG called out in 2005 [1] So what, exactly, can a person substantively say against /this/ piece? I would argue: practically nothing. The best points raised by others relate to fusion in general, which we could have had without this submarine PR piece. 1. http://www.paulgraham.com/submarine.html http://www.paulgraham.com/submarine.html

tutanchamun 9y ago

> The techniques being used are largely well understood science and the downsides of previous approaches like ITER are explicitly avoided. I'm a layman when it comes to these things, but does that mean they have improved tokamak's or will future nuclear fusion reactors more look like the stellarator used in Wendelstein 7-X for example?

dekhn 9y ago

Actually, you can easilly discount this, since they haven't run the experiment yet. Until that happens it should be fully discounted. Happy to pay attention when the experiment is run.