Alpha3031

Biomed is already too complex to simulate fully, which is why simulating is just, not really a thing that is done to find new knowledge. Heck, some quantum interactions are already (so far) computationally intractable, which is why quantum chemistry exists as a field at all, it's far from trivial. That doesn't mean the best practices developed for medicine won't help in the social sciences as well.

The issue with degrees of freedom can be mostly pinned down with preregistration, among other measures. It won't solve the problem of having more type I errors at any specific error rate the more science you do, but it does give assurances the target type I error rate is accurate.

Which specific social sciences do you believe are doing better?

Are we talking about the October 2022 report by Senate Republicans or the April 2023 report from, again, Senate Republicans? Because as I understand it, the official political explanation according to Republicans of a lot of things doesn't exactly comport with reality, so I'm not sure why you believe we should trust them here, given the track record. Unless the idea was to push a particular point of view.

Central banks do not set policy based on what they see inflation is in the past, they make projections of what inflation will be in the future.

Ugh, you know, I think what I hate most about the market fundies is that their perception of what interest rates would be is absolutely untethered from any semblance of empirical reality. Like, sure, it would be fucking nice if the long-run equilibrium rate were anywhere near where they think it would be, but believing r* is close to what it was back in the 70s is a bit like believing you can will the economy into having higher productivity growth through the magic power of wishful thinking. But nooo, it's definitely not having enough free market that's the problem, we should just finish what Reagan started, certainly the union busting had no contributions to the declining labour share at all.

It would be funny if it wasn't fucking terrifying.

Labour productivity in economics is output per labour hour, it doesn't matter what is doing the work. As an extreme example, if robots took over and human labour is no longer required for anything then (human) labour productivity would be infinite, it doesn't matter if no humans are actually doing any labour.

DACCCS is currently insufficiently mature to estimate how "reversible" any specific level of warming is, but current policies are expected to limit warming to 2 to 3 °C by 2100, compared to over 4 °C where no policies were implemented. Net zero emissions by 2050 would even be compatible with 1.5 °C, and that is well within what is achievable with current technology. Plus, with AR6, we have the first estimate of the zero emissions commitment over 50 years, and it is centred around zero, with a 90% CI of roughly ±0.3 °C, meaning it is likely (with "low confidence", meaning this is our first time doing it as part of CMIP) that changes in 20 year average temperatures will be quite small after complete cessation of emissions. That means assuming we ended up on a trajectory like 2 °C by 2100, if we wanted to reverse GSAT at least we would have a better than 66% chance of having another 5 decades of research and development and deployment (plus the 7 decades it took to get there), so once we get to zero emissions it becomes much less urgent.

There will still be permanent changes to the biosphere, but there is no reason to feel doomed.

> Today’s infrared thermometers for fevers have a 2F accuracy. You’re telling me that you can calculate temperatures from around the world pre-1950s and trust that it’s consistent and accurate?

Yes. Believe it or not, the average of multiple calibrated measurements is more accurate than some random off the shelf consumer electronics. My laptop display has a delta E of about 20 out of the box. If someone calibrates it they could bring it down to 5, most likely. I have no trouble believing an actually decent panel can achieve less than 3.

I can probably finagle a way of measuring haemoglobin with my kitchen equipment and some reagents I order off eBay. I might even get one sig fig out of that. That doesn't mean the literal bottom shelf crap I use is going to be the absolute limit of possible precision.

I mean, what kind of argument even is that? "I can't make similar measurements using the equipment I have at home, therefore climate scientists must have either no idea what error bars are or are lying"?

> Which means even with the large amounts of snow that will build-up in the winter, there are always cars in the lot - clearing that is a big job and not sure how to do it with solar panels.

I'm not sure how much sense it'll make in this specific case, but in general, interestingly, with a high albedo (e.g. snow covered or white for some other reason) ground surface, vertically mounted bifacial panels produce very close to equator facing panels with the appropriate tilt. The value of that generation might actually be greater, as it obviously spreads out the profile towards the evening peaks.

For the regular horizontal/tilted ground mounts, NY is quite far north, so you're looking at a mount angle of 43° or so to optimise overall power production, and you can go about 15° more tilt and only lose maybe 1 or 2% of summer production, but increase production in winter when the sun is angled lower. If you go for like 55° to 58° tilt, there's probably not going to be very much snow accumulating on the panels themselves.

> What dimension is this measuring?

Twenty bucks says wafer size, since 300 mm wafers are standard.

Edit: Also, here's the press release for the announcement:

https://www.nist.gov/news-events/news/2024/02/biden-harris-a...

Look mate, your car has always been oxidising. What do you think is more achievable, putting out the fire I set or just adapting?

What? No, of course you're not allowed to try more than one.

Honestly all those non-arson factors that cause oxidation means it's pretty brain dead to assume me setting a fire had anything to do with it. You're probably just going to have to live with all your stuff being on fire, just a hazard of living in a oxygen-having environment.

> Pumped storage is a fine technology, but is there a lot of build potential left for it?

"A lot" is probably subjective, but the two best known global estimates are Hunt et al. (2020) [1] from IIASA and Stocks et al. from ANU re100 (who incidentally also have some interactive maps [3]) which with different cost targets put potential at 17.3 and 23 PWh respectively, which works out to about 2 MWh per person. For comparison, for the past decade, the US ahs consumed about 13 MWh of electricity per person per year, down from a peak of slightly under 14 in 2000 and 2005. With very high levels of electrification, that could potentially rise to 24 to 28 MWh per person per year, or 8 or 9 PWh/yr for the whole country. Total primary energy use is a lot higher, around 90 MWh per person–year or 30 PWh total, this is because both not everything could be practically electrified and the things that could easily be electrified tend to be much more efficient when done electrically. Energy efficiency is also usually assumed to increase slightly in general.

The US specifically is actually above the world average at about 4.5 MWh per capita according to the ANU team's estimates. That's 1.5 PWh per year roughly. In any case, I would expect that there is likely to very likely sufficient potential in most (if not all) grids for pumped hydro to be a significant part of medium duration energy storage (if not all of it), though whether it actually would depends on the costs of other technologies as well.

[1]: https://doi.org/10.1038/s41467-020-14555-y

[2]: https://doi.org/10.1016/j.joule.2020.11.015

[3]: https://re100.eng.anu.edu.au/

Not necessarily true at this point for new projects. IRENA indicates wind projects commissioned globally in 2022 had a weighted average LCOE of $33 USD/MWh. Hydropower projects over the last decade or so (2010–2021) were around $39 to $48. (2022 average was $61, but obviously recent market conditions were far from normal)

Some password managers do offer the option of challenge-response from a hardware key, but technically speaking the password manager vault file can be considered "something you have" so long as you store it securely, like your SSH private key.

> I'm not sure how this could be avoided in the US system of 2 senators per state. You can't exactly send 2/3ds of a senator to Washington if you had a proportional system.

Make the state legislatures proportional, impose a deadlock-free contingency procedure and repeal the seventeenth amendment. Either that or random ballot.

Though, I wouldn't necessarily consider either of those good ways of solving that problem, which would probably involve increasing the size of the senate as well somehow.

The paper states that this is indeed the case, and also cites previous proposals along similar lines:

> Academic researchers have drawn inspiration from the cake-cutting problem to propose processes for fairer redistricting. While this logic has inspired several redistricting proposals (Alexeev and Mixon 2019; Benadè, Procaccia, and Tucker-Foltz 2023; Brams 2020; Ely 2019; Landau, Reid, and Yershov 2009; Pegden, Procaccia, and Yu 2017), none has addressed implementation across the full range of U.S. states and electoral contexts using real-world data. The complexity of previous proposals also precludes real- world applications or even simulation using modern computing techniques.

You missed a zero with the 1921 figure, but it is an interesting question to consider: what exactly is the optimal size of a national assembly? Does it scale best to the square root or cube root of population? Or some other exponent? What are the constant factors? Do larger multi-regional blocs like the US and EU require different styles of representation, and if so does that change the scaling factors? The US could probably do with at least 600 though, minimum.

Abolishing political parties is not possible, people form factions and coalitions based on common interests, and those coalitions form coalitions in order to achieve their goals. You can have a multi-party system, a two party system or a de facto one party system, the difference is that for the latter two, the final coalitions happen beforehand and are presented to voters as a fait accompli.

Strictly speaking, the fundamental complexity classes (in order: L, NL, P, NP, PSPACE, EXPTIME, NEXPTIME, EXPSPACE) apply only to decision problems, which pose yes/no questions. For example, it's commonly said that the Traveling Salesman Problem is NP-complete, but this only applies to the decision version of the problem: "Given graph G with edges weighted with positive integers and a maximum weight n, does there exist a Hamiltonian cycle through the vertices such that the sum of the weights of the edges used is at most n?" In addition, for a problem to be in NP, the polynomial-time-verifiable certificate only has to exist for yes instances, there is no need for it to also be verifiable that no such cycle exists (a problem verifiable in polynomial time for the no cases is in co-NP).

Decision problems are much easier to reason about, which is why they are used to construct and define the fundamental complexity classes, though it is certainly possible to define analogous classes for different types of problems. See for example, Krentel (1988) "The complexity of optimization problems", which considers TSP-OPT: https://doi.org/10.1016/0022-0000(88)90039-6

Fortunately for EU regulators, US judicial precedent is not binding there while their Digital Markets Act is. We'll see if they take this lying down.

Seems weird to include SMIC and not Samsung or Intel. Those two are behind TSMC, sure, but not so far behind that it's sensible to leave their facilities off the map of "makes chips on their own". They're not GloFo.

It's possible to manage a dirty float without setting fire to the domestic economy, you know... In fact, investors tend to be hesitant to invest in economies that are on fire even with favourable exchange rates.

A leftward shift of LM is still a leftward shift of LM.

For some context on the costs, SEC analysis for 2017–2020 puts the median figure at $750k, and the mean is over double, $1.8 million.[1] In more recent years, the costs have only grown. Despite being theoretically possible to run one for less than $10k if you basically just provide the minimum legally required notice, in practice anything like that is very unlikely to actually succeed, and median (mean) costs have been $1.1m ($1.2m) in 2022 and $1.0m ($2.7m).[2] That's just the costs directly involved in proxy fights themselves, there are also often other things that you might like to do to improve chances of success.

Of course, even with the high costs, there are notable successes, such as Exxon having a quarter of their board replaced in 2021. And union organising is not mutually exclusive, in fact, Starbucks unions are seeking three seats on the board this year. But they're notable because of how rare they are, and the difficulties will mean that it's likely to remain relatively rare.

On the other hand, having the precedent set that it's even at all possible does probably make it more likely for directors to acquiesce to even non-binding resolutions.

[1]: https://corpgov.law.harvard.edu/2022/05/25/the-cost-of-proxy...

[2]: https://www.kirkland.com/publications/kirkland-manda-update/...

> IP protection is not cartelization / price fixing.

Dear god, which rock have you been living under if you're somehow not even able to conceive of the similarities with the other industries where monopolies are good, actually. Coase did good work but clearly he was a disaster to public education, if you're not even able to at least recapitulate his arguments. Wait, wait, you do at least know what a marginal cost is, right?

> You responded to me by inventing a strawman argument

Ah yes, the strawman of microfoundations. I really wonder how some heterodox economists manage to even take themselves seriously sometimes. I assure you that having microfoundations for your model is much stronger than not having any foundations whatsoever.

> Your particiaption should be self-evidently valuable if you want to be considered a participant in a debate.

See, pigeon chess. There's no point in debating you lot, I mean Adam fucking Smith, the commie bastard he is, wrote a book about that mercantilist BS, so if three centuries hasn't been long enough for it to sink in nothing I say or do will.

> Incresing internal wealth is always predicated on importation of foreign currency. This is why sanctions work to impoverish nations, in spite of possession of valuable items like oil.

Like, wow, this is some breathtakingly novel take on the paradox of plenty right here, we should get this published so you can win the next Sveriges Riksbank Prize.

There are proposed microfoundations for efficiency wages that do not necessary imply a significant level of bargaining power for an individual worker.

Yeah, 7945HX laptops are going to use a dedicated GPU like the 7900M or 4090 because the integrated graphics on there is mostly... vestigial. 2 CU basically means 2 cores (from the last generation) in Apple terms. Rumours are AMD might shove a (laptop) 4070-ish GPU with 40 CU into the next refresh (if I had to guess the numbering, 9950HS) but I don't think they would see a point in going higher than that (if they even try that much), because the higher GPU performance segments are going to be served by their dedicated GPUs paired with the HX CPUs that have 2 CU.

AMD doesn't really have the volume to outbid Apple on the leading nodes, so what they do get is going to keep going to server/HPC. That stuff is more money for them anyway. Adding more cores, both CPU and GPU and shrinking to TSMC's latest node, it isn't easy but it's technically simple enough that they could do it in a few months if they thought there was enough of a market for it. There just isn't, unfortunately.

> It sucks but apple laptop cpus and mobile cpus dominate anything else out there. I wish there were more competition. And their margins are massive.

Ryzen 7940HS and 7945HX are fairly close to the M3 16 core, if they had actually been able to build them on the same node, and if AMD were willing to shell out for the same die size, it's likely that they could have achieved or exceeded the uplift.

Tax base is based on total income, or total production (, monopolies extract economic profit (contrast, accounting profit) by artificially constraining quantity supplied such that the price is higher than the marginal cost of production. In no world does this increase production in the domestic market. If you want to price fix on only the export market, well, you can encourage cartelisation for export only, sure. Only way to turn that into revenue is to tax it, and politicians don't tend to like putting tariffs on their countries exports, I can't imagine why at all. (There are both good reasons and bad reasons).

> If you are so sure of you're reasoning, then why the rhetorical framing?

Because I'm not putting in more effort if my counterpart refuses to. I'm not going to dignify the "economically globalist society" part with a response unless I'm forced to, I'm still not sure you sincerely believe that, but I can't reason someone out of a position they weren't reasoned into.

That's a reach, but if the primary purpose of monopolies in your view is to fund the military, then there's no issue with taxing them more or even nationalising them, is there?