[icapybara]

I'd rather hear a projection from an engineer/scientist/operations person in the industry. This kinda reads like it's written by an armchair expert who thinks about batteries a lot, but doesn't have much to do with building that future being described.

Sometimes the technical details matter and projected scaling trends aren't an inevitability.

[tux3]

It does seem to be written by someone who's very far above the ground — even managing to throw the blockchain in there at the end.

But the point they're making is reasonable. Just because the author isn't deeply technical doesn't mean they can't fit an exponential and extrapolate correctly.

Exponential growth always has to stop somewhere, but that's not in and of itself a reason to think this year is the year that it will. The napkin math about sodium and battery cost is at least reasonable, it's worth considering seriously rather than handwaving the author away as not an engineer.

[icapybara]

Fair. I guess I have an issue with technology projections that assume the technology will follow some fit, because it always has. Every bit of progress is made with tons of risky work and breakthroughs, and none of it is guaranteed like you would think it is just by looking at a fit.

[XorNot]

> they can't fit an exponential and extrapolate correctly.

Anyone fitting an exponential isn't extrapolating correctly pretty much by definition. As you note:

> Exponential growth always has to stop somewhere, but that's not in and of itself a reason to think this year is the year that it will.

This is a god of the gaps argument. There's no reason it should stop this year, there's also no reason it shouldn't. Fitting the curve is only useful if you're actually presenting an argument as to why for the relevant interval it should continue.

[Dylan16807]

There's plenty of reasons it shouldn't end this year. Go look at the battery sizes that we need to switch halfway to electric cars and to stabilize the electrical grid.

Looking at what happens if growth continues until we get into that range is quite reasonable.

[iSnow]

>Fitting the curve is only useful if you're actually presenting an argument as to why for the relevant interval it should continue.

Which he actually does by looking at sodium batteries.

[mcswell]

> projected scaling trends aren't an inevitability

Reminds me of a projection I read back in the early 1960s (I think). The author charted the rise in speed of human beings over ten or twenty thousand years, where that speed had increased when horses were tamed, clipper ships were built, steam trains invented, automobiles, airplanes, and then rockets. (Assuming this was just after Gagarin, that got "us" to 5 miles per second.)

He pointed out that the acceleration was (ahem) accelerating, with thousands of years between humans running and horses being domesticated, vs. about sixty years between the Wright brothers and Gagarin. Extrapolating, it was clear we would exceed the speed of light (using a warp drive or something) by the year 2000.

Of course the current record speed was set in 1968 at about seven miles per second, and not even equaled since 1972. So much for extrapolation.

[jtbayly]

He claims the expert estimates have been wrong every year (too conservative).

[marcosdumay]

To the extent that have been expert estimates out there, they do have been consistently wrong. The same happens to solar and wind generation.

But well, I haven't seen any that don't have a conflict of interest into claiming fossil fuels will continue to be required. And that's a large part of the problem: you just won't find uninterested experts publishing estimates.

[XorNot]

Well the other issue though is that people looking for predictions want conservative predictions because they're investing. Over-estimate and you lose money, under-estimate and you leave money on the table but don't trade away future possible gains.

[marcosdumay]

Hum... People investing in renewables and batteries do not use predictions. Those have at most some 3 years of building time, usually a few months.

People investing in fossil fuel plants need predictions, and despite they wanting to see conservative numbers, that bias means complete doom for them. They need the opposite bias if they want to survive.

Things get a lot more complex once you start to look at components industries. But then, it's not clear they use predictions for anything.

[jillesvangurp]

And sometimes an exponential is staring you in your face and you just don't realize it. This has happened before. Early computer scientist did not imagine anything like you and I take for granted and put in our pockets without thinking about it every day. That's only a generation or so ago. Two if you are half my age (50).

IMHO, the theme of this century is making cheap, sustainable energy so ridiculously abundant that we'll be wondering what the hell we were doing before and how we managed without it. There are so many technological breakthroughs converging on making that happen that IMHO this is just going to happen. It's a question of when, not if. The timelines are uncertain, but not really. The author of this article is extrapolating a few trends over a time scale that is rather short. He could be wrong. Even by a factor 5. And it would still happen on a reasonable timeline. And I don't think he's going to be that far of the mark. 2030-2035 it will be RIP ice engines and fossil fuels. You'd be out of your mind to use anything else than dirt cheap electrons stored in dirt cheap batteries. At 50$ per kwh, it's a no brainer. At 5$/kwh, you'd have to be bat shit crazy to use anything else. That's 'only' a 10x improvement.

Assuming all innovation grinds to a halt in 2024 and that no technical progress will happen beyond 2024 seems like the naive point of view when there's so much happening that is well funded and seemingly on track to get some kind of results. The opposite view on this is of course that progress is a foregone conclusion. Some things will taper off and other things we haven't even thought off might pick up the slack. Between now and 2030, you can make a few educated guesses though. Which is what this author is doing.

Anyway, cheap, clean energy is transformative. Most of the major challenges right now are directly or indirectly bottle necked on energy. Making energy cheaper matters. 2x is nice. 10x is nicer. 100x is what we might actually see in a few decades. Anything in between would be transformative. Anything beyond that is hard to imagine but yet not unlikely. We might actually nail fusion at some point. Who knows? It might even become cheap to do it.

But we have a nice fusion plant that we orbit around beaming down orders of magnitude more energy than we actually need. We're learning how to harvest it using solar panels; a trick plants and trees have of course mastered ages ago. This article is about leveraging batteries for storage. The two things combined are a thing of beauty.

The point about sodium ion is that there are no exotic/scarce materials in there. The materials are cheap. And we're not going to run out of them. How many twh. of battery could we need. Tens, hunders, thousands? We only use about 25pwh per year worth of electricity right now. That number is going to go up of course. What would you do with 25000 twh of battery? Annual production is about to cross the 1twh/year. And most of these batteries last a few decades. 25pwh of charged batteries is a lot of power. And yet we might have that sitting around in a few decades.