I had thought about this fairly lately. Seemed suspicious to me that the universe seemed to be composed of tiny tiles, which it is, if I am understanding Planck's constant correctly. This makes things seem very like the design of a video game, where it is possible to lay out all the tiles and calculate them sequentially and then assemble the grid. It also seems suspicious that there is no evidence of other technological species. It occurred to me that, if one wanted to give the simulated species complete free reign, you would have to allow computational power for it to create multiple simulations of the evolution of the universe it found itself in. Each one of these could eat up quite a lot of computer time, but if one limited things to one species, you might be able to manage things. It would also allow one to render the universe to perfect detail from the one point of view, but allow the rendered resolution to be reduced the farther you looked from this view point. I expected we would be living in the social model of a grad student. If he were a professor, he would have more computer power available.

Everyone seems to treat this as some sort of legal question. If that is all that this is, then Shamir's argument makes sense. No one seems to deal with the issue of encryption and putting a back door into an encryption algorithm. The NSA supposedly did this with the elliptic curve methods, and now they evidently are untrustworthy and unuseable. The RSA method involves picking two large prime numbers and keeping them secret. A back door might involve pretending to pick large primes, but actually picking smaller primes so as to make it easier to decrypt. This would break the alogrithm for everyone. There really isn't any way to put a back door into mathematics. This case seems to be something that the Maker and root your iPhone people should already have solved. Basically, dump the data on your phone to an image file on disk and bang on it until you start seeing intelligible strings. Apple doesn't need to be involved at all.

It's good to see the nefarious plans of evil genius succeed! Lennart Poettering has succeeded in taking total control of all linux implementations! .... ne hah hah...(evil laugh).

I'm seeing something like this: As an empire reaches its terminal state, its elites become entrenched with closed in conflicts between themselves which are beyond understanding to those outside. All intellectual endeavours become increasingly captured by these internal conflicts, because in order to succeed, one must win support of one or another entrenched faction. Eventually, all endeavours are completely consumed in the the struggle between the factions, and no productive intellectual endeavour takes place. This sort of thing happened once every couple of centuries in China, and it seems as though America is entering something of the same sort of era.

The tech industry didn't really invent this. I think it probably started when MBA's were put in charge of running companies and no one was promoted to management through the ranks. From there things proceeded to the huge layoffs of engineers that happened whenever government decided it was tired of the space program or missile defence. No attempt was made to retrain anybody through any of this. Every ten years a great crisis was announced, we were suffering from a terrible shortage of engineers, then a few years later, large numbers were laid off. Doubtless many individuals could actually have been retrained for various specialities since they had a large base of knowledge in math and engineering, acquired at considerable public expense, in some cases. The people doing the hiring, however, did not understand anything at all about the positions they were hiring for. They were in personnel because it had nothing to do with math, among other reasons, so they weren't able to assess whether people were qualified and re-trainable or not. Things have just gone on like this, decade after decade, ever since.

I had been using Google docs for school papers because it let me edit them both on school computers and at home and was good enough. I stopped using it when I started using Libre Office for most things and found I apparently could not load documents created in in Libre Office into Google, no matter what format I saved them into. They would load and allow viewing, but would not allow editing. I assumed this was just Google's way of avoiding problems with MIcrosoft that it didn't see any point in fighting over. Because of this, I just switched to saving everything onto dropbox, and avoided Google completely.

So now all we need to do is correlate how long it takes for each person to click a link after seeing an unpleasant image, and we know who to hit up for cash for the NRA! Life is getting so simple!

Having just completed a Coursera course on machine learning I must be an expert, but the neural networks described there were essentially just a large matrix of parameters. One propogated the data forward through the matrix and made a prediction of the class of the object, then propogated the errors backward and corrected the parameters of the matrix accordingly. This method basically draws a lot of small line segments around each class, so indeed, it will always be possible to place individual points near a class boundary, but on the wrong side. The description of a neural network in the olfactory bulb which I read about years ago worked completely differently. The excitatory and inhibitory neurons acting on one another produced a quasi-periodic signal. This was basically described as noise, but the amplitude of the signal over the olfactory bulb encoded a signal. So the odor of wood chips produced a particular pattern of amplitude over the olfactory bulb that was different than that for any other odor. If the animal was shocked when exposed to the odor of wood chips, then the pattern of amplitude changed, so an association between different stimuli was already being made at the lowest level. I can't quite tell, but it doesn't seem that a simple matrix will be a good model for this.

This present change in technology represents a fundamental shift, something at least comparable to the change at the beginning of the industrial revolution, possibly even larger. Technology of the sort represented by Watson can replace all the employees of a call center. It isn't a question of minimum wage or jobs being created elsewhere. If there is a question which can be answered by searching through somewhat structure information in reply to a spoken question, then it can almost be answered by a computer at this present moment. If software is modularized and all interfaces are carefully controlled, then it can almost be written by a computer at this present moment. There seem to be only two human functions which definitely cannot be replaced: jobs in which human interaction is what is being sold, and consumerism.

This isn't simply a threat to some labor group or collection of individuals, it is a complete reorganization of society that is underway, without any real awareness. Owners of companies continue to make money, while employees are gradually replaced. Even technology workers will be increasingly replaced as code production is automated. Office workers are replaced by programs resembling IBM's Watson. The only jobs that cannot be replaced are those that involve selling actual human interaction. This indicates an economy with owners at the top and a giant hole in the middle which no one can cross.

Microsoft is in a position where it's core market is about to be superseded by mobile technology. They thought that they could do something like they did with Internet Explorer and just make their main competition part of the operating system, while making Windows 8 all touch, so that what you use in the office is what you use on the phone, and tablet. This approach has not worked for them so far to move into this already established market. People seem more willing to learn new interfaces than they were a few years ago, and the interfaces are just better and more intuitive and reliable, so the fear factor and desire to keep to the familiar is working against them rather than for them. They need to give these things away to businesses, with designed integration with business software and processes, so that the Surface can become the device that businesses just assume they need for mobile applications. Either that, or Microsoft will become a server operating system company.

EFI or something like it really is necessary at some point. Real mode must go away, and BIOS depends on real mode. Intel processors can't go on pretending to be supercharged 8086's forever. EFI allows much more evidently as well. A whole operating system can be crammed into it for embedded systems, or phones and tablets. I was very worried when I bought a laptop recently and realized that I would need to do a EFI install, since it seemed bricking would be a real possibility, but the current generation of Ubuntu LTS worked completely painlessly. So I would say the basic linux shim method is a painless klug for the moment from the user perspective.