Hacker News Reader

Comment by JumpCrisscross | Hacker News Reader

Eye opener is the chart at 18:03: "from 1970 to 1995, [fusion] power and energy increased much faster than computing power (Moore's Law)" [1].

Magnetic-confinement fusion stalled, according to Dr. Whyte, because to get more power (and efficiency) we need stronger fields. To get stronger fields we needed bigger magnets. To support bigger magnets one needs bigger structures. Big magnets and big structures will absorb more magnetic and neutron flux; that increases maintenance frequency, complexity and cost.

Better superconductors let us get stronger fields without bigger magnets. That changes the cost function.

[1] https://youtu.be/KkpqA8yG9T4?t=18m3s

View on HN

"In 2011, Japanese scientists stumbled across a discovery which increased a metal compound's superconductivity by immersing iron-based compounds in hot alcoholic beverages such as red wine.[40][41] Earlier reports indicated that excess Fe is the cause of the bicollinear antiferromagnetic order and is not in favor of superconductivity. Further investigation revealed that weak acid has the ability to deintercalate the excess Fe from the interlayer sites."

I guess iron based superconductors are a relatively new and hot area of research.

https://en.wikipedia.org/wiki/Iron-based_superconductor

The more I read about this, the more I kept waiting for a Douglas Adams joke. Science is just silly sometimes.

The lack of investment in this field is absolutely ludicrous given the potential payoff. I suspect some amount of lobbying by entrenched power and fuel interests that would be rendered obsolete, but I think more blame can be placed on that awful meme that brainwashed a generation of scientists and engineers into dismissing fusion as a boondoggle:

"Fusion energy is 20 years away and always will be."

This shows how much damage a superficial catchy meme can do if it manages to slip in and override more nuanced and informed thinking. It's why I distrust catchy sloganish rules in other fields too, such as "premature optimization is the root of all evil" (leads to erooM's law in software) or "never roll your own cryptography" (discourages people from learning about how to create secure systems), etc.

exactly, these slogan almost look like a doctrination campaign. but this is off topic.

20 years ago they said "Fusion energy is 50 years away and always will be." Soon enough it's going to drop to "Fusion energy is 5 years away and always will be." Which is just silly.

We'll still be able to confidently say Virgin Galactic is only 6 months away from sending up paying passengers.

Right. The light at the end of the tunnel recedes until it doesn't.

Your post hit a lot of notes I agree with. Programming dogma doesn't make a lick of sense sometimes. I wonder if the "don't implement cryptography yourself" rule is disingenuous, and really about not necessitating more work from our governmental APTs to get peepholes into everything. I'm also amazed at how often people cry premature optimization when there's potential to reduce hardware cost by orders of magnitude by not writing slow, memory hungry junk by default.

I feel you. While there's plenty of history (hundreds of years) to indicate you shouldn't rely on rolling your own (though Jefferson did, and did a helluvagood job.) It's crazy easy to fool oneself about how great one's own crypto is. However, what you can do is roll your own and then encrypt your data once again - after that initial encryption - with excellent standard methods. This can allow decryption in parallel in some circumstances - if you want the best speed - by encrypting a one-time-pad with the roll-your-own and sending that in advance (and decrypting that pad in advance, so you just have to xor it into the mix later.) When encrypting fer real, first use the one time pad on the data and then encrypt it with regular methods.

Reply on news.ycombinator.com