|
> Yes, Fonar v GE is a ridiculous one when generalized, and I do wholeheartedly agree that the "enablement" requirement is very weak currently and a ton of patents don't meet it in my eyes. Take the PageRank patent, for one, often cited as an example of a good patent. It is very lacking in implementation details, so much so that a blogger set out to implement it and ended up with a bunch of posts complaining about all the undue experimentation he had to do. Glad to hear it. > [4] actually finds EDT is better than others with respect to plaintiff win-rates. I might be missing something, but I don't see it taking the really high settlement rate into account. If you include those as mostly losses, well yes, Northern California still sucks as well and the overall win rate sucks. > There is very little "widespread" re-invention going on Using SSL with RC4, scan to email, pick any "we own the internet" patent, really. > The order in which you "interpret" or execute those bytes gives you a completely different program. No, the hardware performs an instruction loop no matter what is in memory or the registers (unless you break it). > Now consider that you fill that memory up with programs that generate other programs, or variations of themselves. Something like, say, Conway's game of life. I cannot begin to calculate how many different programs that could generate. None of which fit into the device (you've already used up all memory, remember?). It's also a finite number. You're thinking of a Turing Machine with infinite memory (wish we had one, but we don't). There are functions with finite values that are not computable incidentally (the Busy Beaver function, which incidentally is related to your idea, is one such function). Yes, there are more programs than fit into memory for any finite amount. Sadly, we do not have infinite memory. Yes, you can increase the number of possible programs by adding memory (this should be no surprise to anyone who has ever used a computer and needed more space to install X). Every program (equivalently, every mathematical statement) can be identified with a number. Even this text is nothing more than a very long number. I believe I already mentioned Godel's work. > no different from the number of ways physical objects can be arranged. That's infinite, though (as far as physics knows). Anything you can fit into memory will be finite (though it can be extremely large, as you've noticed). One is exhaustible and enumerable, the other is not. The computer is designed to let you put any value whatsoever into its memory. We do not have a general atom-arranger that can make any construct out of matter. Though how I wish we did have one of those from Star Trek! ... I'm sure you'd then have people patent making X with the synthesizer... for every X that already exists. Which is about where we are with the computer, no? That's why we think that new synthesizers (computers) should get patents but not every trivial, already existing X that you can imagine putting on there. At least when it's a new X we can respect it a bit, but when you're doing old X + computer, it gets silly fast, because many, many people can figure out how to do old X + computer ... but the lawyers think it patentable for some reason. > The computer's properties are not altered?!? It goes from being a heap of semiconductors to actually doing something! How is that different from a machine going from a heap of components to an useful implement? It goes from executing an instruction loop on no data... executing an instruction loop on data. Yes, the instruction loop was a very fundamental, useful idea, precisely because it can compute any algorithm given appropriate input. The computer is unlike any other invention precisely because we found such a general mathematical construct to put behind it. But the data is just a large number. Yes, printing a few GB of data as a decimal number would run you out of paper, but there's absolutely no reason we couldn't do it. |