Chapter one starts with a fictional example. Say you have been trying to develop an algorithm at work that validates designs for new products. After much work you haven't found anything better than exhaustive search, which is too slow.

You don't want to tell your boss "I can't find an efficient algorithm. I guess I'm just too dumb".

What you'd like to do is prove that the problem is inherently intractable, so you could confidently tell your boss "I can't find an efficient algorithm, because no such algorithm is possible!".

Unfortunately, the authors note, proving intractability is also often very hard. Even the best theoreticians have been stymied trying to prove commonly encountered hard problems are intractable. That's where the theory of NP-completeness comes in:

> However, having read this book, you have discovered something almost as good. The theory of NP-completeness provides many straightforward techniques for proving that a given problem is “just as hard” as a large number of other problems that are widely recognized as being difficult and that have been confounding the experts for years.

Using the techniques from the book you prove the problem is NP-complete. Then you can go to your boss and announce "I can't find an efficient algorithm, but neither can all these famous people". The authors note that at the very least this informs your boss that it won't do any good to fire you and hire another algorithms expert. They go on:

> Of course, our own bosses would frown upon our writing this book if its sole purpose was to protect the jobs of algorithm designers. Indeed, discovering that a problem is NP-complete is usually just the beginning of work on that problem.

...

> However, the knowledge that it is NP-complete does provide valuable information about what lines of approach have the potential of being most productive. Certainly the search for an efficient, exact algorithm should be accorded low priority. It is now more appropriate to concentrate on other, less ambitious, approaches. For example, you might look for efficient algorithms that solve various special cases of the general problem. You might look for algorithms that, though not guaranteed to run quickly, seem likely to do so most of the time. Or you might even relax the problem somewhat, looking for a fast algorithm that merely finds designs that meet most of the component specifications. In short, the primary application of the theory of NP-completeness is to assist algorithm designers in directing their problem-solving efforts toward those approaches that have the greatest likelihood of leading to useful algorithms.

0. There are tiny people who are not invisible but there is something that makes it so humans ignore them. That specific mushroom interfered with whatever that is so that we can see them.

An argument for #0 and in favor of #2 is the #2 requires the mushroom to cause an hallucination that includes a lot of generated detail to be realistic and consistent with how an actual tiny person would behave and affect the world.

#0 on the other hand is just turning something off.

An argument for #0 over #1 is that #0 does not require actual invisibility. If there were actually invisible tiny people we still probably would have discovered by now the physical mechanism that can make things invisible, which w e have not. Unless we are going to consider magic or psychic powers or some other such thing completely beyond known physics we can almost certainly rule out actual invisibility.

The definition is in in the bill, but the copies I have seen are formatted in ways that make copy/paste too much work, and a quick comparison looks like it is almost identical to New York's law targeting such feeds (except Illinois numbers there list 1-8 whereas NY letters it a-h), and I happen to have that in an easily pastable form, so here is the NY definition:

> "Addictive feed" shall mean a website, online service, online application, or mobile application, or a portion thereof, in which multiple pieces of media generated or shared by users of a website, online service, online application, or mobile application, either concurrently or sequentially, are recommended, selected, or prioritized for display to a user based, in whole or in part, on information associated with the user or the user's device, unless any of the following conditions are met, alone or in combination with one another:

> (a) the recommendation, prioritization, or selection is based on information that is not persistently associated with the user or user's device, and does not concern the user's previous interactions with media generated or shared by other users;

> (b) the recommendation, prioritization, or selection is based on user-selected privacy or accessibility settings, or technical information concerning the user's device;

> (c) the user expressly and unambiguously requested the specific media, media by the author, creator, or poster of media the user has subscribed to, or media shared by users to a page or group the user has subscribed to, provided that the media is not recommended, selected, or prioritized for display based, in whole or in part, on other information associated with the user or the user's device that is not otherwise permissible under this subdivision;

> (d) the user expressly and unambiguously requested that specific media, media by a specified author, creator, or poster of media the user has subscribed to, or media shared by users to a page or group the user has subscribed to pursuant to paragraph (c) of this subdivision, be blocked, prioritized or deprioritized for display, provided that the media is not recommended, selected, or prioritized for display based, in whole or in part, on other information associated with the user or the user's device that is not otherwise permissible under this subdivision;

> (e) the media are direct and private communications;

> (f) the media are recommended, selected, or prioritized only in response to a specific search inquiry by the user;

(> g) the media recommended, selected, or prioritized for display is exclusively next in a pre-existing sequence from the same author, creator, poster, or source; or

> (h) the recommendation, prioritization, or selection is necessary to comply with the provisions of this article and any regulations promulgated pursuant to this article.

Trump 2 is a breakdown of the separation of powers and checks and balances that were supposed to give the US continuity. With that no longer functioning any deal you make with the US is really a deal with the current President which future Presidents may or may not honor.

It doesn't matter how reasonable the next President is because just electing a good President doesn't fix the underlying system. If other countries don't see that fixed so that another Trump cannot happen they will not consider the US to be reliable for long term things.

However, Caltech and MIT have a reason for this that probably doesn't apply to a large (35000 undergraduate) public university like UM.

Caltech, MIT, and UM all have incoming classes that did very well in high school. Around 88% of the UM class was in the top 10% of high school. It is 96+% for Caltech and MIT.

At all three most students will be going from having been one of the top students at every school they have attended to a school where they are more average. This drop on average will be bigger at Caltech and MIT because their students are clustered more tightly toward the top of their high school class.

Another thing that makes the drop bigger at Caltech (and MIT?) is that UM has a way more flexible schedule. For example if you take physics in your first year, they have I believe three different courses for that. One if for people majoring in life sciences and emphasizes the physics most useful in those areas. Another is for engineers and others doing practical applications. It is more rigorous. The third is for physics majors.

Compare to Caltech. First, there is no "if you take physics". It is a first year requirement. Second, there is just one course. Go to Caltech to major in English (which does occasionally happen! [1]) and your first year you are taking the same required physics and math courses that people there for physics and math are taking. Third, because admissions specifically looked for people with very strong math/science ability, the professors don't have to slow down. Those courses cram a lot into a short time compared to even the courses meant for people majoring in their subject at less specialized schools.

The net effect is that for a top high school student Caltech and MIT are much bigger academic shocks than UM because even though at UM you can find comparable courses structural differences due to its size and its much bigger number of students mean it doesn't make you take a bunch of rigorous courses right from the start at an insane pace.

[1] It is rare. The reasons some have done it include:

• They intend to be science writers or science journalists.

• They intend to go to medical school. Medical school admissions committees like English majors and they like a hard science background.

• They intend to go to law school. An English major with a serious hard science education and a law degree is a good fit for patent law or for going into science policy work.

A summary for those not familiar with it:

• Operating systems on devices whose primary user is a child must provide a way for the parent to provide age bracket information about the child.

• They must also provide an API that apps and app stores can use to find out that age range if they need to have limits on what children can do with them.

• They amended it to not apply to most open source operating systems.

Note that there is no actual age verification. It just uses what the parents put in. It is really just requiring each OS to have a standardized parental control system.

Put some (or all) of your circuits on a transfer switch or generator interlock and connect your solar system to that.

Then legally your system is the same as a backup generator install. It should have permitting and inspection requirements similar to those of adding a circuit. That will just a city permit and a city inspection.

That's likely a big part of it.

AI data centers need an order of magnitude or more power than a cloud data center, and a lot more are being built.

Before with the lower power requirements and with slow and steady growth a data center project was handled by the utilities just like they handled a new factory or a new large shopping mall. The utility might need to add substation of some local transmission lines but it wasn't a big deal.

An AI data center uses so much more power that even if the utility has enough capacity overall to cover it, the AI data center needs that all at one site meaning quit a bit more new infrastructure needs to be built.

Now add in the number being built. The utility now is much less likely to have generation capacity to cover them, meaning the utility will have to buy electricity on the open market which is more expensive.

Another factor is that many AI data centers cannot wait for that and so find sources of energy other than the grid to tide them over for the several years it might take before the grid is ready for them. These sources are often not good for their neighbors.

It was the MLAT thing that the various civil liberties groups object to (I'll cover the problems with those down below). There was very little objection to the first part.

The first part was not controversial because pretty much every country has something equivalent (for reasons I'll cover below), as did the US except specifically in the case of data covered by the SCA due to poor drafting.

One of the big reasons for the SCA was created was the emerging "third party doctrine" meant that instead of having to get a warrant or subpoena against you to get your data they could simply subpoena it from any of your service providers that had it. The SCA made it so the third party doctrine subpoenas would not apply to stored communications.

There were still cases where the government would need to compel the service provider to turn over the data. They wanted something with the probable cause requirements of a warrant but the delivery method of a subpoena. (A subpoena asks someone who controls the data to turn a copy over. A warrant is for when the government wants to raid the data center and seize the data. Since that involves the government directly acting where the data is located it only applies to someplace where they have jurisdiction).

So they created a new thing, the SCA warrant. The called it a "warrant" because it had the probable cause requirements of a warrant, but neglected to add something saying that in other respects it functions like a subpoena. I'll call this a pseudo-warrant.

The SCA was not the first pseudo-warrant. That would be the warrants under the Wiretap Act of 1968. Territoriality questions did not arise under that because by its nature the data it sought copies of was always in the US.

With the SCA the data might not necessarily be in the US. Years later Microsoft argued that because it is a "warrant" it should have the territorial restrictions that normal warrants have. The CLOUD Act clarified that it was indeed supposed to be like a subpoena as far as territoriality goes.

There have been some more pseudo-warrants created since then, but their drafters learned from the SCA and made sure the original legislation was clear on just what they were.

The reason pretty much every country has something like that, going back well before online documents, is because not having such a thing leads to big problems. If anyone in the country could shield documents from subpoenas (or whatever the equivalent is called in that country) by merely storing them across a border every company with documents that it needs to keep but that might be incriminating later would get sent to a storage facility across a border as soon as they were no longer actively using them.

For example as soon as a car company in Detroit releases a new car all the documents where during development engineers brought up safety concerns which management decided to not address would be sent across the bridge to a storage facility in Canada.

With electronic documents it is even easier. You would not have to wait until you aren't actively using the documents to stick them outside the country. Just stick your file server across a border and make sure you only have copies in country when someone is actively reading or editing them.

And so pretty much everywhere subpoenas compel someone in the country who controls the documents to fetch them (or copies) and turn them over. The actual location of the documents is completely irrelevant.

The thing that was worrying about the CLOUD Act was the MLAT provisions. MLATs are treaties where the participating countries agree on law enforcement. They include things like sharing information and cooperating on investigations. Normally these are enacted just like any other treaty. The executive branch negotiates them and then the Senate votes on ratification.

The CLOUD Act adds an expedited process where the Attorney General and the Secretary of State can sign an MLAT. Congress is not involved. These agreements allow foreign law enforcement to make requests directly to US service providers instead of going through the diplomatic channels normal MLAT requests go through, and they allow them access to stored communications that the SCA would normally block.

There are some safeguards. The foreign government is not supposed to intentionally target US people who are in the US and are not not supposed to use the data they get to infringe freedom of expression. There's also a 180 day window before these executive MLATs take effect during which Congress can block them by passing a joint resolution to do so.

Civil rights groups and many others were not impressed with those safeguards.

Would it be feasible for people to flush their toilets with bottled water for a week?

When combined coal and gas provide about 62% of China's electricity and 58% of US electricity. (Sources differ slightly on percentages. When you add in all fossil fuel electricity sources, China and the US get about the same percentage of electricity from them).

It is this way because China has plenty of coal and very little natural gas, whereas the US thanks to fracking has a lot of natural gas and it is much cheaper than coal so since then gas has rapidly replace coal.

That's the static picture. Now look at the first derivative. China is not just massively ahead at building renewables, they are massively ahead at building support infrastructure.

Look at UHV transmission lines for example: China has 60 000 km, quite a bit ahead of the 0 km in the US.

Anyone happen to know if it is true for countably infinite sets?

Given N points, N > 2, can you arrange them in a Euclidean plane so that (1) they are not all on the same line, and (2) every line that goes through two of the points must also go through at least one more of the points?

The theorem says that you cannot do this.

Wrapping is fine on some material, but when there are tables it can be very annoying.

But someday many people might need wheelchairs that include some functionality to help with obstacles and are affordable. Those primitive impractical machines engineers build now are where they figure out much of what will go into those later machines.