You'd be amazed how much effort it takes to jump through all those hoops. I've seen ERP systems with fairly complex configurations to keep track of all the shell games.

When the transaction is A pays B, B ships product to A, the systems are pretty easy to build/maintain. When the transaction involves money and physical goods going through completely different paths it gets really messy staying on top of it all.

Companies do it because it still pays off, but as with most of the finance sector this stuff is just a drain on the economy. If we could get rid of it all we'd be much better off as a society.

Maybe they just want to have a chance at being able to read the page titles in the tab list, which is impossible when you have 47 tabs open?

I don't get the problem with opt-in push notifications any more than I get the problem with opt-in desktop notifications. They allow browser applications to do stuff that non-browser applications are used for all the time.

You'll need to enable these for them to work, unlike on phones where they tend to be enabled by default. However, on android you can completely suppress the ability of an application to display notifications from its settings page.

Git is one of those tools which you have to grok to really be effective with. The data model in the tool is not intuitive, but it works really well in practice. Once you understand how it actually works, then the rest falls into place.

Now, something git really suffers from is inconsistency in its command-line tools. Specifying a branch name varies in syntax from tool to tool, and there are a million other cases like this. So, even if you understand what you need git to do, it can take a lot of digging to figure out the right way to do it.

Not to start a flamewar, but I find systemd to be in a similar boat. It has what amounts to a class inheritance aspect to it with units, and targets are often poorly understood. There are tons of "instead of this type that" guides out there, but I think that if you want to effectively use systemd you have to grok its data model.

A Turing machine is a mathematical construction. You're trying to use the halting problem as a rationale for the universe being non-deterministic. However, the halting problem only applies to Turing machines. A Turing machine can't contain a black box, because that isn't part of the definition of a Turing machine.

It is a bit like proving that there are a countable number of integers and then trying to say that there must be a countable number of irrational numbers by just redefining the meaning of "integer." You're playing word games, but that doesn't prove anything.

To be honest, I'm not really sure how you could prove anything about the universe using an argument purely from discrete mathematics, unless it is a proof that there is no such thing as determinism at all (which certainly would be an interesting claim for a mathematician to make). Math is a world all its own, and while it can be used to describe the physical world, it has an existence apart from it in a sense.

Again, you're playing word games. Determinism requires that the universe has some state, and some set of rules that determines what its next state will be (which is a really rough way of putting it when time isn't discrete and is relative, but I don't think we're arguing about that). If you're arguing that determinism means something else, then we're just talking past each other.

Honestly, the only reason I'm continuing this discussion is because I thought you might have an interesting argument for determinism based on the halting problem, which was your original point. I wasn't sure if you were just having trouble communicating your ideas, or if they were not established in rigor.

I don't really care whether I convince you. I don't intend to be disagreeable for its own sake. I'm just skeptical. If you're going to assert that the universe is non-deterministic that is a really bold statement, and I'm not going to simply accept it at face value.

The only stand I'm taking is that I don't believe anybody has shown conclusively whether the universe is deterministic or not. I'm open to arguments one way or the other. I'm fine with thought experiments and hand-waving arguments, but I'm not going to accept them as some kind of conclusive proof.

The problem with such an approach is that your machine is part of the universe, and therefore must modify it as part of performing its evaluation of the future state of the universe. The machine must also be larger than the universe to do this, since it needs to maintain in its memory the entire current state of the universe and some number of future states to perform the calculation.

Such a machine cannot exist. The fact that it can't exist doesn't mean the universe isn't deterministic. It just means that you can't write down the state of the entire universe using only the matter present inside of it.

You can't build a machine that perfectly models itself for the same reason that you can't build a box that can fit itself inside unmodified.

And Turing machines execute the instructions they're told to execute. They don't figure out what they'd do and then do the opposite, unless you just invert the programming. A hand-waving argument about doing such a thing doesn't prove anything about the universe.

I have no idea whether the universe is deterministic. I'm just pointing out that neither does anybody else.

What does that even mean? The instruction table says that if the condition is A, move left, if the condition is B, move left, if the condition is C, move right, and so on. A Turing machine operates entirely in the now (just like any other conventional computer). It doesn't have to guess what it might do - it just looks at the current state, looks it up in a table, and does what it says.

If you're saying that it should try to figure out what it will do 10 steps in the future, then that is basically the halting problem and it is not possible. Nor is it necessary for its behavior to be deterministic. The only thing determinism requires is that given a current state there is only one correct immediate future state. That is exactly what a Turing machine is - a mathematical construct that is perfectly deterministic.

Actually, they still don't wan to be. They argued to another court that the FCC didn't have the right to make them a common carrier.

I think their goal was to get this case dismissed since they're a common carrier, and then get the controls of being a common carrier tossed since they aren't one.

Turing machines only support a few instructions. It can move left, move right, stay in the same place, change state, and write something in the current position.

There is no instruction called "terminate when the function does the opposite of whatever the halting machine says the function will do." I don't even know what that means. Turing machines don't say anything - they just execute instructions.

The behavior of a Turing machine is completely deterministic. If the instruction table says to move left given the current conditions, it will move left.

I'll accept that definition of the universe, but recognizing that "all deterministic systems" might be an empty set.

You're basically trying a proof by contradiction here, I believe. However, your wording is really loose.

First, what do you mean by "predict the outcome of the experiment?" What experiment?

Then you say that you "should be able to contrive" an "experiment" that can predict the outcome of the "experiment," but that it will have the "output" that is the opposite of whatever was predicted. What does this statement even mean, since it appears self-contradictory, like saying let the set A contain the number 5 and not contain the number 5. I can't tell whether your conclusions follow from your premise as your premise seems confusing at best.

It seems like you're trying to do something like this: Let f(x) = x + 3. So, f(5) = 8. So let's redefine f(x) = x + 3 if x5, and it equals 9 if x=5. You haven't made a contradiction, you've just swapped one deterministic function out with another.

You also brought up the halting problem, and it isn't obvious to me how that relates. The halting problem has nothing to do with determinism, just predictability. The behavior of a halting machine is completely deterministic - it has a finitely definable starting state, and given that starting state will always end up executing the same series of steps. However, it is unpredictable in the sense that you can't tell what the result will be with certainty without actually running through all the steps to get there.

Sure, I'll buy that. However, in practice if the universe contains elements that are non-deterministic, it seems extremely likely that the physical construction of a computer would not be entirely free of those elements.

Sure, but how do you specify which three have to collaborate for any particular domain, and who do you have to trust to have made that certification?

Simple, stick the certs in the DNSSEC records. Then only registrars between you and the root can spoof you. If you don't trust the USA, then pick a registrar in a country you do trust, and now the USA can't spoof your records.

If you want convenience you'll always have to trust somebody, but with the DNSSEC proposal only a few companies could spoof any particular website (with the list being different for each website). The Chinese government couldn't spoof nsa.gov, and the NSA couldn't spoof government.cn.

The root still involves some challenges, but that is high-profile and doesn't need to change much, so there are a lot of options there which won't scale up to the entire DNS tree.

The machine is still deterministic. You can predict with 100% certainty that if the condition on the evaluation is true, it will do A, and if not it will do B. You can even determine what the condition will be by running all the steps of the program until you get to that statement. That is determinism.

The only way you could have a "computer" that wasn't deterministic would be if its operation were influenced by randomness (true randomness - not algorithmic pseudo-randomness). Such a device would not be a Turing Machine.

In fact, whether it is possible to really build a finite Turing Machine depends on whether the universe is actually deterministic. Any computer exists in the physical world, and is subject to things like cosmic rays that can impact logic decisions apart from the content of memory, or for that matter change the content of memory apart from the program design, etc. So, if the cosmic rays are non-deterministic then it is not possible to build true a finite Turing Machine. (By "finite" I mean a machine with only limited tape length - a true Turing machine requires limitless memory.)

The existence of the Halting Problem REQUIRES determinism. The Halting machine itself is deterministic. Given the current state of the machine, you can perfectly predict the next step of the machine. You just can't predict whether it will ever finish running without walking through the steps until it stops (assuming it does, and you'll never know if you ran it long enough).

Determinism is more about knowledge of the current state and the rules the game operates by.

Sure, but look at the list of charges in the summary. The guy could have gotten somebody shot, but the charges are all about computer crimes and whatever, probably because those were the most serious laws that they could get him for breaking.