Straw

Most modern RNGs should be faster than memory bandwidth (when optimized), so unless your list is small enough to fit in cache, its unclear if this is faster?

Market systems work especially well when supply or demand are constrained- they allocate the scarce resource more effectively!

The data center demands are known years in advance. Its not like they magically appear out of nowhere.

Perhaps we should eliminate the utilities guaranteed monopoly then! Monopolies are usually dysfunctional.

There isn't an externality here in the traditional sense. No one has a right to purchase anything at any particular price. Should people be disallowed from bidding in auctions because it increases the price of the item in question?

No, people who live in geography absolutely should not control the land surrounding them. That's effectively land hoarding, and incentivizes deeply counterproductive NIMBYism. People should primarily control the land they own.

Why should they? If demand for X goes up, and price goes up, the whole point of the price system is to incentivize all users of X to use less and producers of X to make more.

Here's an example where the benefit is obvious:

Say I set up a desalination plant for a city with water shortages. I use a lot of electricity; this drives up electricity prices. But I also sell the more scarce water, lowering the price of water. If I make more selling the water, then to first order I'm helping people overall- they value the water more than the electricity. Perhaps I put other, less efficient desalination plants out of use, or simply other uses of electricity that people value less than the water I provide.

Although this is a neat idea, the author appears to be confused about how generative modelling works.

They repeatedly claim that previous approaches rely on factorization to reduce things to guessable chunks to avoid the 'blur problem'. This is a misunderstanding. Previous approaches solve this problem by modelling a distribution as output rather than a point. Factorization is one way of representing the distribution, but the key point is that even for the small chunks we predict an output distribution, not a point estimate.

Factored models have no issue producing TV static as an image. Each individual guess is impossible, and yet they don't produce a constant grey averaged image.

Normalizing flows, though not considered particularly efficient, are not factored at all, do not proceed in small steps, and yet still have no issues with blur because they predict a distribution.

Their approach appears to be a hard version of a latent variable model. It may be a good idea, but it isn't a fundamental change in the way they suggest.

Sure. The most straightforward example would be utilities in much of the US. Companies like PG&E are literally granted a monopoly in a region.

Another example would be epipens. There is a competitor, but unlike most drugs, if the doctor prescribes you an epipen, the pharmacy can't give you a generic substitute, it must be name brand.

Tesla is being heavily protected from Chinese competitors by US auto tariffs.

Credit Ratings agencies (in large part responsible for 2008 crash) have direct government granted status.

Taxi medallions, Jones act, sugar tariffs.

These are the ones that come to mind right away, but there's a whole slew of stuff- any regulation that's easier for a big business to comply with than a small competitor, for example. Even worse if the big business manages to influence the regulation to enhance this effect. Personally I'd guess Boeing is an example of this; yes, aviation does have high capital costs but I suspect that FAA regulation + Boeing cozy relationship makes it much harder for new entrants. Of course, this isn't to say the FAA regulation isn't helpful or doesn't improve air safety, just that it also has this effect.

Grocery store profits aren't larger than any other industry. Nor are they larger than in the past. Large grocery store chains do not have high profit margins. Sure, the total profit may be large if you run 1000 grocery stores.

What do you mean by they are the wealthiest in the nation? None of the richest people in the world built their fortunes on groceries?

Price fixing is a phenomenon. Anti trust is a policy agenda. What entity would carry out the proposed agenda, when the parties supposedly fixing are in different countries?

How long did the fixing agreement last successfuly? DRAM has famously gone through booms and busts where supply outstripped demand and led to very cheap prices and no profits.

OPEC worked about as well as any cartel. Could you give an example of a cartel maintaining their position without heavy government help? Even the drug cartels in mexico, with extreme willingness to coerce, cannot maintain a monopoly and have competed drug prices in the US down heavily! If you consider memory a cartel, then it has only remained so by... drastically improving the quality and reducing the cost of its product? Hardly a problem.

There are many other factors affecting the size of corporations. For example, a significant increase in the amount of regulation in other sectors of the economy.

More importantly, the raw number of corporations isn't how you measure a monopoly. Profit margins are the key sign of monopoly pricing- but in practice, corporate profit margins have been relatively stable in percentage terms over time. Prices for virtually all consumer goods are way down in real terms over the last 30 years.

In particular, trade might get you more competition with fewer corporations, but via competition with overseas producers. These might show up as a few importing corporations, representing a whole slew of competitors abroad.

I'd say exactly the same about antitrust. Can you give an example of a lasting monopoly not backed by some government policy? At any point in the last 150 years?

You claim that a law that prevents or hinders you from making transactions with certain parties, while privileging other parties by exempting them from these restrictions, doesn't make it easier for the privileged parties to form a monopoly?

Groceries are an excellent example of competition. Profit margins on groceries are razor thin. The large chain leverage results in lower consumer prices. Large grocery chains don't have anomalously high profits, ROI, or really anything that would indicate harm to consumers. Groceries are cheaper than ever compared to median wages.

The memory market is international, its not clear what anti trust even means here. Even so, look at the price of memory over time, say the last 20-30 years. Adjust for inflation and its even more extreme. Sure, it is up recently with AI demand, but the general trend is drastically reducing cost and increasing quality. Memory technologies are evolving quickly and it's a capital-intensive business, so its not surprising that there are a few leaders. There's no guarantee of instant competition in free markets.

Cartels rarely last long without enforcement. There's such a strong incentive to defect and secretly sell more under the table. Even OPEC could not maintain one. Memory right now is expensive simply because demand outstrips supply and there is significant capital investment and lead time to increasing supply.

If you wanted to do the most to improve competitiveness possible, by far the most effective strategy would be free trade. Trade barriers prevent competition from overseas, making monopolies much easier to maintain.

No, governments tend to produce monopolies.

Can you find true monopolies in the free market, in the last 100 years, that have lasted for at least 20 years, without some kind of government backing/suppression of competitors? I only know of one.

Things only count as monopolies if they have monopoly pricing power; Google is not a monopoly, they cannot even charge 1 cent per search.

In practice, despite economies of scale, large companies age and become bloated, inflexible, etc. So there's typically plenty of room for new companies to compete.

If a monopolist buys all upstarts, then that creates an even stronger incentive to make competing businesses! You don't even have to succeed in the market, you just have to get bought up! The more they buy out, there stronger incentive there will be to enter the market. No one can do this indefinitely- except with some kind of government support- say a regulation that makes it difficult to enter the market.

Oh, I wouldn't say that. The summary for policymakers they produce is incredibly alarmist compared to what their actual reports predict.