It's about 8 light-minutes away.

In all seriousness, the combination of (solar + wind + hydro), battery storage, and long-distance grid connections is all we will need. It's cost-competitive with fossil fuels today and it's only getting more attractive as tech improves.

The best place for fusion is in the Sun's core, where any stray neutrons it generates are intercepted before reaching anything we humans care about. This fusion powers solar, wind, and hydro energy Earth-wide.

The cost of renewably generating power is attractively low today and will be even lower in the next few years. The only real challenge left is to find a way to store and distribute the energy widely, robustly, and flexibly enough.

I'm betting on high-voltage DC cables, which can link unsynchronized grids and don't dissipate RF radiation if they go through salt water. Their range also scales with the square of their voltage*. Intercontinental, multi-1000s-of-km power cables are science fiction today, but there's a smooth path to their adoption (e.g. linking Great Brittan to the Sahara before a trans-Atlantic cable).

I also think it's high time to let the price of electricity float. Electric car chargers (and many industrial and HVAC applications) can be set to gorge on power when it's cheap and plentiful and turn off when it's scarce. Storing money is much easier than storing energy. Free markets work well for sending price signals. We shouldn't try to reinvent the wheel and make a planned economy when it comes to power**.

We're all nerds and love the idea of a fusion reactor as a magic bullet. Me too! However, we shouldn't let a hopeful dream get in the way of seeing that clean power is already within our grasp.

---------

* For fixed power, if you up the voltage 10x, the current goes down 10x, so the V=IR drop goes down 10x. You then lose 1/10th the voltage out of a 10x baseline, thus a 10x voltage increase cuts Ohmic losses on a fixed power transfer by 99%. Another way to slice it is that increasing the voltage 10x increases the acceptable transmission range (i.e. the range at which you lose a fixed proportion of your power to resistance) by 100x.

** Back when power was largely generated by fossil fuels, a planned power economy made more sense since costs were level and generation in any geographical area was a natural monopoly. Today, true costs vary with the wind and sun, and it would helpful to let the market do its thing whereby the grid acts as the arbiter between power generators and consumers.

A friend of mine is doing a hardware startup with 2 people that he says would have taken at least 3 extra coders pre-AI. He and his partner are doing the valuable EE and mechanical parts of the business. A lot of the software they need is not what's differentiating their startup, and AI is just fine. For them, it sounds like AI is giving a 150% productivity gain, and without it their business idea would only be marginally viable.

Sad story! Try talking over your situation with an LLM to see if leaving really would be impossible. It's not going to be easy for sure, but doing hard things is a great way of generating interesting life stories.

The current chapter in your life sounds bleak. It may be time to turn the page.

Uncomfortable, beautifully shot, nuanced, and at times both hilarious and horrifying. It's a contender for "best movie of all time" in my books.

Agreed! Perhaps power bills will be a little above $zero, but the combination of solar, batteries, and long-distance transmission is already cheaper than fossil fuels in many circumstances. And this trio is only going to get cheaper and better. I don't foresee a bright future for grid power coming from either fossil fuels or nuclear, unless insane laws prop them up.

Thanks, ToasterMonkey. That's exactly what I meant.

Unless we know for sure that P != NP, we can never know for sure that any encryption is safe. (If verifying that a secret crypto key is correct is in P, and P = NP, then finding that crypto key has to also be in P.)

However, we currently know that there's a way for a big enough quantum computer to crack today's RSA encryption. It's generally supposed that several state-level agencies (and probably some other actors) are storing today's encrypted messages on the assumption that they will be decryptable in the future. Why not add protection against future interlopers now? Personally, I want all of my sensitive communications that rely on RSA to be irrelevant and obsolete long before the first big quantum computers turn on.

Today we don't know if it will be 10 or 100 years before there's a big quantum computer. Given that there's at least some chance it will be less than 30 years, and seeing how slowly we transition away from old software, I fully support a careful augmentation of today's standards with something that at least has a chance of being quantum-resistant.

(My opinion also is that quantum computers in practice are going to be significantly faster than classical computers only of a pretty tiny set of tasks. RSA was a little unlucky that arguably the best quantum algorithm can break it.)

I disagree. I want there to be low-friction ways to buy tiny parts of large companies. A stock exchange is waaaaay better than every investor independently evaluating every potential company. No system is perfect, but the SEC and stock exchanges are MUCH better for the little guys than the law of the jungle.

I know you're asking an anti-science user about what they would want. That said, even the left-wing scientists didn't act optimally.

One thing we could have done is be more morally flexible about challenge trials. See https://www.1daysooner.org/ for a missed opportunity.

If the MRNA vaccines had come 6 months sooner there might have been a chance to preempt a lot of the COVID mutations that made vaccines less effective. It was physically (but not politically) possible to have a timeline that fast.

My take is that at root, it was moral realism that made COVID such a mess. "Moral realism" is the idea that there *is* one set of simple rules we can stick by that will always tell us the right thing to do. I am not a moral realist, and I think the world is complicated enough to send us occasional circumstances where it's better to set aside the moral codes that were appropriate in the past.

Because of moral realism we decided it would not be ethical to allow volunteers to be deliberately exposed to COVID in vaccine challenge trials, nor to commander the know-how on making MRNA vaccines at scale.

Right-wingers often complain about the vaccine not being a choice. I am miffed that I was forbidden to choose to try out the vaccine candidate back in February 2020, all because we collectively adopted the idea that the correct moral code for all circumstances could ever be as simple as "first, do no harm".

My friend next door found a loophole to the "no-AI-therapy" rules. He runs an LLM bot that does Tarot card readings, but first it asks you a bunch of questions about your life. It can then legally say something like: "Oh, the next card is the High Priestess. It looks like you may be in an abusive relationship. Want to talk about it?"

My $0.02 is that AI-based therapy might be very useful to a segment of the population who can't afford human-provided services. Don't let perfect be the enemy of good.

Companies are valued based on the latest stock transactions. If I can sell one billionth of MSFT for $4,000, then MSFT is valued at $4 trillion. That's just extrapolating the company's whole value based on the latest marginal transactions.

To buy all of MSFT would take more than $4 trillion (since there would be true-believer holdouts demanding more) and if all owners were to try to sell all of MSFT they would get less than $4 trillion for it (since the owners are dumping the stock). Still, for buying and selling small fractions of the company, $4 trillion is exactly the right valuation for the whole of it.

You're not wrong. Still, wouldn't only the crooked candidates be the ones taking the fall?

Some physicists these days like Sean Carroll think that the arrow of time and causality arise thermodynamically. All fundamental physical laws are time-reversible, so the only way for there to be a difference between "forward" and "backward" in time is because one direction was closer to the Big Bang than the other. Increases of entropy are the engine of time asymmetry and causation; there's no other way for us to sensibly say the past causes the future (and not vice versa) since at a fundamental level the physical laws all work as well forward and backward (caveats about the weak nuclear force notwithstanding; it's not the weak force that prevents you from un-scrambling an egg by swishing in the reverse direction).

The current best picture implies a temporally upside-down world "before" the Big Bang by our coordinate system, which is precisely as problematic as an American saying that by their coordinate system, in Australia they build their roofs "under" their houses. Both branches of the timeline have arrows of time pointing away from the shared Big Bang just like both continents have "up" pointing in opposite directions from a shared center of the Earth.

If the idea of the Big Bang being an entropy minimum holds, then saying black holes from a previous universe "cause" dark matter now is a problem. That's because it defies the idea that the Big Bang was a causal root of everything that happens in our local space. To make the idea in TFA work there would have to be another route to an arrow of time. While that's not impossible, it's an extremely tall order, and it would require a lot of rethinking of fundamental concepts that underpin our ideas about how the world works.

Some topical videos:
Up and Atom video on the arrow of time (great intro; 15 minutes: https://www.youtube.com/watch?...)
Sean Carroll's latest thoughts (more depth, cutting edge, > 1 hr: https://www.youtube.com/watch?...)

Agreed!

Einstein taught us that gravity fields are indistinguishable from acceleration. Suppose you have an ideal acceleration sensor. If Earth's local gravity field is unknown at the 0.1% level, you'll have an error in your acceleration signal of about 1 cm /s /s. Let that accumulate for an hour and you are 1/2 (a) (t^2) = 65 km away from where you thought you were.

Much of that will be in the vertical direction, but not all of it!

If you're a public company you'd have to declare a big expense like paying a ransom.