Veedrac

I've suggested a different take on PNBL in the context of gambling before, to protect people against addiction without outright making it illegal as is widely done today. The idea is if you want to gamble more than some unprotected quantity, you have to first allocate the money and leave it for a month or two. People with insufficient impulse control or financial security will find that hard, but light recreational users shouldn't be inconvenienced too much.

The same ideas can work for other things, though it's easiest to apply to predatory behaviour.

Fun fact, you can extend this logic to measure the upper limit for how discriminating your RNG's test is, by shuffling a 2^n long vector with a CSRPNG and testing that until failure the same way. When I tried this ~8y ago, I believe these tested about 2 bits better on PractRand than PCG for 32 bit outputs. Sadly it's not really plausible to run this algorithm for 64 bit outputs.

The main thing this knowledge is worth when designing RNGs is that it tells you how reasonable it is to 'lose' N bits. Eg. if you're 2 bits worse than PCG, you're about twice as much worse than ideal.

That New York has made all other bike share companies illegal doesn't mean Citi Bike, a private company without monetary subsidy, isn't selling into a market.

It's free advice but if you're a part of government and think a service is good and should be encouraged, I recommend not proposing bills whose only contents is making the existing market for that thing illegal.

So much of policy success comes down to doing obvious and reasonable things, and most of the problem is how to incentivize making those choices. For all China's flaws, they've figured this out.

The best news here is that we might finally have a prosaic means to escape our modern-era applied biotech stagnation, the same way solar has appeared as a means to kick the feet out of traditional energy sources. China is pretty new to being an R&D powerhouse, but there are few more worthy causes.

I don't think you're understanding the comparison here. The Chongqing metro area in 2015 had a population of 13.4m, today it's 18.2m, a 4.8m difference, or 36% growth.

The Netherlands went from 16.9 to 18.1 today, a 1.2m difference or 7.1% growth. Good by nobody-builds-anything regulatory paralysis standards. Standstill by Chongqing standards.

These articles which are little more than blind scarequotes peppered with ad hominems invite little more meaningful discussion than more ad hominems. What is the value of a comments section filled with little but "this is stupid and anyone who believes it is a snake oil salesman"?

They contain zero-order holds, but they also contain reconstruction filters. Ideal reconstruction filters have infinite width, which makes them clearly different in kind. And most DACs nowadays are Delta-Sigma, so the zero-order hold is on a binary value at a much higher frequency than the source signal. The dissimilarities matter.

A pixel is simply not a point sample. A camera does not take point sample snapshots, it integrates lightfall over little rectangular areas. A modern display does not reconstruct an image the way a DAC reconstructs sounds, they render little rectangles of light, generally with visible XY edges.

The paper's claim applies at least somewhat sensibly to CRTs, but one mustn't imagine the voltage interpolation and shadow masking a CRT does corresponds meaningfully to how modern displays work... and even for CRTs it was never actually correct to claim that pixels were point samples.

It is pretty reasonable in the modern day to say that an idealized pixel is a little square. A lot of graphics operates under this simplifying assumption, and it works better than most things in practice.

Storage matters for widespread direct-consumer use of hydrogen, but in the example I gave it's not a big deal: just put the electrolyzer next to the solar panels. You only need to store ~12h of hydrogen production, and burn it onsite.

A more realistic world won't be implementing the Dumbest Possible Refutation, and would overbuild solar less than this in the first place. In that case you do care a lot more about storage, and that's a large part of why I suggested ‘synthetic fuels powering existing fossil fuel plants’ would be a saner strategy. But what exactly that world looks like is in the details, and not critical to the broad point I was making.

Hydrogen generation systems exist on the same balance. When the input electricity is expensive, you want to build them to be more efficient, and that costs money, in catalysts and low-loss reaction chambers and such. If a huge amount of energy at peak times is free, then the optimal point is very different, and indeed if you try to minimize capital costs you end up needing something barely more sophisticated than a kettle. Kettles aren't particularly expensive to run!

While competition will quickly drive this towards a more even balance, as cheap storage displaces yet-more excess solar buildout, the point of the argument was just to show why naïvely extrapolating to extreme overproduction (>2x) is misleading.

We don't yet live in a world where electricity is reliably overproduced by 2x. Renewables are ~100% of the US's new electricity production, but still only a small fraction of its total electricity production.

The author misses a perhaps unintuitive point: the cost of storage depends also on the cost of energy. By the time you've overbuilt 2x, a full extra 100% of your demand is sitting around literally free at odd hours.

Traditionally, moving energy around means batteries, and yes maybe your battery costs more than just generating new electricity from a less efficient new solar panel at odd hours. But batteries are optimized for energy being expensive, where losses are wasteful.

Consider this really simple, dirt cheap alternative: plug your free energy into a pool of water and collect the hydrogen from it. Burn the hydrogen later, and point the light at your idle solar panels. It's hellishly inefficient, but I repeat: the energy is free. You are only minimizing capital costs, at least until other people catch up and start shifting load some other way.

The sane point on this curve probably looks something along the lines of a mix of batteries and synthetic fuels powering existing fossil fuel plants. The nice thing about going all the way to synthetic fuels and not hydrogen is that long term storage becomes trivially cheap, so it starts offsetting your winter load as well.

> Lots of people predict it

Yup.

2009 https://news.ycombinator.com/item?id=855684

2017 https://news.ycombinator.com/item?id=14790251

2018 https://news.ycombinator.com/item?id=17184054

2019 https://news.ycombinator.com/item?id=21613997

2022 https://news.ycombinator.com/item?id=30622300

2023 https://news.ycombinator.com/item?id=35091237

2024 https://news.ycombinator.com/item?id=40739431

The author of the technique has an explanation of why screen-alined dithering is hard in their FAQ. https://github.com/runevision/Dither3D/discussions/12

Particular, it's hard to deal with anisotropy, and naive approaches to fix this hurt either break the operation as a dither with uniform dot sizes, or struggle to avoid becoming regular at angles.

I agree with the claim, though, that you lose quite a lot by having the dither pattern stuck to surfaces. There is a charm to it for sure, but a high quality screen space dither would be cool. And so I experimented with some ideas and came up with one. Forgive the jank, I'm learning as I go.

https://www.shadertoy.com/view/WfBGzK

Note that this is using a random dither versus a patterned one, which I think hurts the aesthetics a lot, but I am certain this is incidental to the concept. I think there are probably much smarter techniques, and I have an idea for a large rewrite already, but those are details, just take this as a proof of concept.

Note that it's black to move and black's pawn moves upwards. All move but one has instant counterplay from white.

Here's the board; you can enable the engine to get the answer: https://lichess.org/analysis/standard/8/6B1/8/8/B7/8/K1pk4/8...

It's not obvious to me that this is worth much over

    (x > 9) + (x > 99) + (x > 999) + (x > 9999)

(x > CONST) does tend to need a pair of instructions to get the status flag into a register, but those can also fuse with the adds. Critical path latency is cmp-setae-sbb-add, or also four cycles.

It's just someone who involves themselves in AI but as a slur.

It's true Falcon 9 has flown for a long time, but its growth rate has been a shockingly consistent 1.6x mass to orbit YoY since the beginning, so at any given time the majority of flights occurred in the previous two years.

The Starship second stage is surprisingly not all that heavy compared to the 78t Shuttle orbiter dry mass, despite the many-fold difference in volume and payload, plus Starship needing to be a legitimate second stage rather than just an orbiter.

Compression and tension are quite different loads. There have been rockets in history that would collapse under their own weight unpressurized. Neutron's second stage is a hung tank for similar reasons. Bucking is a pain. Super Heavy can obviously support its own weight, but tension is always going to be the easier load path.

"The robotaxi has no plug" — Elon Musk

That paper is unserious. It is filled with unjustified assertions, adjectives and emotional appeals, M$-isms like ‘BigTech’, and basic misunderstandings of mathematical theory clearly being sold to a lay audience.

It's not just that we have $25/y now and had ~$24B/y then, to suggest it is practical. Costs are also vastly reduced over those benchmarks. Starship HLS cost the government under $3B. For that same price you could buy 30 Falcon Heavy flights at wholesale rates. New Glenn looks to launch soon and we don't know its price but it should be competitive, and it's a plenty capable ship.

Of course, depending on these as Artemis does now would only improve prices, not timelines, but it's not like commercial options are all a year old. Marginal launch has been approximately free in Artemis terms for a good while now, possibly even back in the pre-SpaceX dark ages when US launch was a monopoly flying on Russian engines with a billion dollar a year protection racket tax.

Ultimately it comes down to most of the money NASA spends on Artemis being spent on rent seeking by large old space companies that know job losses would look politically disastrous, captured in cost-plus contracts with hiring mandates, tech reuse mandates, and financial incentives not to execute. NASA is spending about as much on a mobile launch tower going overbudget on a cost plus contract as they have on Starship HLS. Guess which one had Congresspeople giving empassioned speeches about NASA's unreasonable spending habits!

> Wake me up when a LLM figures out stable fusion or room temperature superconductors.

Man, the goalposts these days.

Thanks, that makes the position much clearer.

Some questions:

You say, “if there's enough pressure to make a little crack, then the fluid can instantly expand and immediately make a big crack”. My admittedly quite surface level view of the research is that it is viewed as feasible in this regime.

“Close to the brittle-ductile transition conditions of pressure and temperature, new findings suggest that fractures are sufficiently permeable to allow fluid circulation and, in case of insufficient fracture density, enhancement strategies are likely to be successful”

https://www.nature.com/articles/s41467-019-12146-0

I also believe that EGS fracture enhancement marginally prefers hydro-shearing (crack expansion) rather than hydro-fracking (crack formation), so if creating cracks is problematic, that leaves options open.

I also note that to my understanding gas fracking is already a thing, cf. nitrogen fracking. So gaseous behavior doesn't obviously seem like an instant write-off to me.

I'm sure that isn't convincing to you, but it's a bit challenging for me as a layman wrt. geothermal to see why I should trust your gut here, so to speak, and I'm wondering if you have a concrete argument I can evaluate on merits?

For example,

"The hypothesis that the brittle–ductile transition (BDT) drastically reduces permeability implies that potentially exploitable geothermal resources (permeability >10−16 m2) consisting of supercritical water could occur only in rocks with unusually high transition temperatures such as basalt. However, tensile fracturing is possible even in ductile rocks, and some permeability–depth relations proposed for the continental crust show no drastic permeability reduction at the BDT."

https://www.nature.com/articles/ngeo2879

I'm definitely not claiming to take these on faith, I'm just saying I haven't really been given a reason to believe those arguments are less trustworthy than your claims otherwise.

cf. "but I wish they finished writing the equations out"

The article mentions EGS but, as far as I could tell, only seemed to present it to contrast it with the claim that Quaise is using a worse single-bore strategy.

If I'm just misreading, and it sounds like you're saying I am?, it'd be really helpful to show your working so I can see where the models are differing. There is a factor 100 difference somewhere, it shouldn't be that hard to spot!

Thanks for the reply!

I'm afraid I'm not really following the argument though. I don't have the technical background to judge how economical EGS is or not, I just want to understand this energy-based argument. I checked to be sure, and Quaise's initial plans (per cofounder Matt Houde) are indeed EGS-based[1], just deeper. It seems correct to me that if they succeeded at building wells that way, they would produce at least comparable energy to standard wells.

It's entirely possible that EGS just doesn't work at really deep depths as you state here, but this seems like a qualitatively different argument to the one presented in the article.

[1] https://youtu.be/yz6rRw59Huw?t=675 "but what we're interested in in Quaise is this novel idea here all the way on the right which we call like to call superhot rock EGS systems"

Actually, they address the energy balance question at the end of that talk.

https://youtu.be/yz6rRw59Huw?t=3016 "[...] we could be using around five megawatts for the drilling process to drill our wells, and let's say we use that five megawatts over a year to drill three holes, so we get an injector and two producers. We predict that configuration of the two producers and an injector at superhot conditions can produce something around 50 to 100 megawatts of electrical energy, again owing to the benefits of producing this superhot, supercritical steam."

They also answer a question on borehole stability, admittedly claiming largely that they don't know.

https://youtu.be/yz6rRw59Huw?t=3254

This is a really interesting perspective but I wish they finished writing the equations out. My attempt at verification didn't match.

Per [1], “[wells] with a regular production casing diameter of 200-250 mm have an average capacity of 5.5 MWe”. Quaise has deep bores that could potentially have higher temperatures, but let's stick with 5 MWe.

At the article's 12 MWh/m³ for drilling, and 250mm bores of depth 10km, WolframAlpha tells me that is 6 GWh.

Divide through, you get 1000 hours or about a month. This doesn't match their “significantly over 10 years” they gave before mentioning waveguide losses.

The main difference I suppose is the thermal conductivity comment, where I didn't follow why Quaise wouldn't be able to use enhanced geothermal approaches. More specifically I think that if a Quaise well has ~1% the energy output of a normal geothermal well, it's pretty weird to frame the problem as about the energy cost of drilling, and not the whole factor-100 reduction in energy output.

To be clear, this seemed like an interesting article and I'm not claiming my napkin math is definitive, I really am neither an expert nor someone who has spent a lot of time investigating this. I do think some more clarity on how the math looks would help their case.

[1] https://www.thinkgeoenergy.com/report-success-of-high-temper...

The problem is dominantly that with traditional techniques the US doesn't have useful geothermal energy resources, so not having geothermal energy historically doesn't mean that newer approaches that aim to work in far more areas will also fail to scale.

Skepticism is reasonable around any new technology, but the arguments for geothermal are convincing enough that it seems easily worth the attempt.