Miscellaneous Operating Systems/Hardware

robespierre wrote: For any (sufficiently powerful) formal system, there will always be a true statement that it can't prove... this has implications for spam classifiers.

A few things to consider. Yes, what you say is correct, but will it practically matter? The goal here is to approach human capability - that is a real thing that exists - so we have an existence proof of a physical thing delivering the kind of classification resolution we need. Might we scratch our heads when we encounter a page that says, "This page is spam". We might, but do we care? I don't think so. In almost all practical situations if we can approximate human capability in this area, we've come pretty close to making this a non issue.

Now, if the human brain isn't merely a physical thing, and there are quantum pixies (ala Penrose), or diktats from the Higher Dimension guiding our spam classification capability, then perhaps we are in fact in trouble.

sgifanatic wrote: A few things to consider. Yes, what you say is correct, but will it practically matter? The goal here is to approach human capability - that is a real thing that exists - so we have an existence proof of a physical thing delivering the kind of classification resolution we need. Might we scratch our heads when we encounter a page that says, "This page is spam". We might, but do we care? I don't think so. In almost all practical situations if we can approximate human capability in this area, we've come pretty close to making this a non issue.

Umm, no offense but you're being silly.

a) spam is not the issue, spam is crap in your mailbox

b) like spam, the issue would be trivial to solve. The underlying problem is that the people who now control the web do not want to solve it. The people who now control the web (google et al) are the problem. This crap is all a by-product of their commercializing bullshit.

c) even worse, the internet authorities (internic ?) also have their heads up their asses. "You have to promise that the name you registered this site under is real." Yeah, right. That works.

I hated this when I first encountered it but 2014 is not 2000. Probably it's not the only way to clean up this mess but it is one way : In China, you have to have a license for a website. The license is free but you have to produce documents to verify who you are. Then you get an ICP number which must be on the front page of the site. People can easily verify you. No phony companies hiding behind "privacy" curtains*. If there's no ICP number or a phony ICP number on the site then the hosting place gets shut down. No ifs ands or buts. Drastic, but it works. And not perfect, but better than nothing. People can trust that a .cn website belongs to a real physical entity whom they can find. Maybe it's a shitty company but at least it's not an invisible thief.

The uncontrolled web is well on its way to becoming nothing more than cable television for the "consumer" but a telescreen for the powers-what-be. This is not a problem which artificial intelligence can solve.


* The whole "privacy" thing with website registrations is total bullshit. So is the entire domain registration process. This stuff was cute in the beginning but at this point, the internet is a public resource and needs to be regulated. Not pseudo-regulated by google, truly regulated for the good of the public.

AI cannot force people to use proper names when they register a website. And I understand the difference between SEO stuffed/robo pages vs mailbox spam. However, a plugin that leverages a text classifier + learns add'l features (e.g. even loosely verified ID tying multiple properties to the same registrant - fake or real) could be used to filter out crap web pages from Google results, as an example.

But please feel free to disagree. I was just sharing my perspective.

sgifanatic wrote: However, a plugin that leverages a text classifier + learns add'l features (e.g. even loosely verified ID tying multiple properties to the same registrant - fake or real) could be used to filter out crap web pages from Google results, as an example.

Make the people who are being roughed up pay more to avoid broken kneecaps ? This is called a protection racket.

It doesn't solve the problem.

From what I've seen of the current Amiga scene, accelerators these days are pretty much all FPGA softcore emulation, then ramping up the clock rates to get the performance and keep compatibility with the 68k series, ppc seems to have died off a bit.

Not sure if FPGA's could cope with a MIPS 10K+ emulation given the difference in architecture?

Not sure about modern MIPS processor compatibility with the older versions used in SGI boxen though.

There is a large enough Amiga scene, especially in Europe, to allow folks to justify that kind of design. I am thinking less complex and more out of the box in the SGI context. Specifically, an ARM board that offers a modern browser over xdmcp via the Ethernet connection. Would also provide firewall, USB and similar services. Add-on would likely be external, but could be made to match aesthetics of popular SGI workstations.

That's not really an accelerator though, and is something most people can do already on their "other" machine(s)?

You could build up a customised FreeBSD/Linux etc image for that if you wanted to make life easier I guess.

You are right, its a customized image and a maybe a nice case.

hardkernel have some pretty nice dual/quad core arm boards that would do as the web browser/ gateway thingy

http://www.hardkernel.com/main/main.php