Saturday, March 21, 2015

A glimmer of intelligence

No, I am not talking about Artificial Intelligence, I mean Andrew Ng as reported here.
Finally, a sane comment as opposed to the loonies worried about comic-book robot nasties taking over.
Maybe, just maybe Stehpen Hawking is good at theoretical mathematics relating to black holes, (those thing that cannot be easily detected because there isn't much there to detect ). . . . but he does seem to have overstated the problem of malignant AI  - but then if you have read any of my previous posts you will know that I can't see any "machine intelligence" appearing soon. [ No, although I am sorely tempted I won't make a snide remark about human intelligence here. ]

All you need to do is look beyond the hype about how fantastically powerful the next generation of computers will be (and it's always "the next generation") and look at the real progress being made and you will see that it is slowing down as we hit some serious barriers, and these are not financial, they are physical barriers to more miniaturisation. Of course, there are still plenty of new or different techniques to try but this means more time, more research and no guarantess that any of those options will pay off.  We have already done most of the cheaper, easier things to squeeze more transistors onto a chip and now each step is harder, more expensive and gaining less.
Once they started talking about "Cloud" computing I suspected that Personal Computers were hitting some sort of limit anyway.

Actually I'm kind of glad since my current PC cost me a heap and if it didn't last a few years more I would be very annoyed.

 Note the button that says "Space Phone". This is a very old remote, not sure what the space phone was.

Here is ashot of new Apple Laptop's innards: the small square top centre is all the mobo there is:
They sure skooshed all the working bits down -hey wait: that looks about the same size as a phone. . .

The wonder of reverse image search

I have known about it for a while now but only recently I have found some use for TINEYE.

Tineye does "reverse image search": what that means is basically that you can give it an image and it will find all of the places that image can be found on the net. Useful? well consider that you have a low res picture and you want a better copy of it. You could try using "de-blurring" software on a zoomed up version of hte one you have but the results are less than great . . . . . or you could try Tineye. 
It can also be useful if you make original images and you want to see if anyone else has been posting them without telling you . . .

And now a public apology to my friend S. who really is innocent and didn't deserve the cruel comment I posted previously. I chalk it up to having my trust clobbered recently by being robbed over the net and some other trust issues that I won't discuss here. Suffice to say that I hope things will be straightened out soon.

And now a small cartoon courtesy of The NIB relating to the dire economic situation in certain nations:

Guess the Titanic already sunk huh?

Tuesday, March 17, 2015

Drone Wars

Drones are everywhere in the media. Personal drones fitted with cameras are getting cheaper, meaning more people will use them to take pictures or even maybe spy on other people,  and this leads in turn to . . . anti-drone drones.   . . . . which will undoubtably be followed by defence drones, and then . . . . and then I'll be waiting nearby to pick up some expensive technotrash I can turn into something more fun.  As if there weren't enough stupid surveillance going on already.
Go here to see Ed Snowden saying "Spying is necessary  - but it needs oversight and should not be conducted against all of us" (paraphrased) - nice idea Ed, but I can't see that happenning soon.
"Snowden pointed out that mass surveillance is “one of the few places in the global political debate where we have a choice.” If you don’t want a dystopian hellhole where governments around the world can see everything you say or do, there’s something you can do about it: use end-to-end encryption."
Uh . . . . can anyone else see the fault in this? Assuming that end-to-end encryption works, We need to ask what exactly we are going to be encrypting, and ask what the hell is the point of it.

Recently I had a mysterious contact who only ever talked to me via email disappear.  I never for a moment thought that this contact was anything other than a scammer trying to get information but I didn't care, mainly because I didn't give the person any useful information as far as I know: in fact, over the time I spent in contact with Mr. "S" I accumulated details that suggested to me who that person was simply because a person's language and knowledge can't help but leak. (Everything leaks, see elsewhere).
For me, that really  the point: Why should I care if some corporate computer knows my favourite flavour or my long-dead pet's name? It won't give anyone even the slightest hint as to passwords or where I keep some valuable item.  Only on a collective, "blah-blah percent of people in South Park like to eat Brand X Chocolate Icecream" type basis will collecting that information have any value.  Yes, I am concerned about huge multinational corporations bullying nations for thie own gain, but the honest truth is I can't think of anything I do online that would require me to encrypt my messages or other data because I just don't do much of that - the only exception to that is money transfers and guess what? that is already protected and encrypted.

If you are dumb enough to give Farcebook or any other online "service" all of your personal data such as your email password, where you went to school, who all of your friends are etc. etc. you will just have to live with the results.  There is no escape, Big Data already has you, running and hiding won't help you any - you still need to live.
Yes, a big pile of S**t may be coming down the turnpike but that is nothing new. We live under the threat of nuclear war right now, climate change is already here and those Big Players want us to ignore it because that might damage their profits. 
When I was a boy, I noticed that outside the classroom boys formed into gangs around the biggest mouths and the biggest fists.  I learned to hang out in the library at lunchtime and that helped both my health and education - but I imagined that when those kids grew up things would be different: how wrong I was. The same tribalism, thuggery and blatant stupidity I saw on the playground just transferred to the business world . . . . .
Sorry folks, enlightenment has been postponed. Come back in another century, maybe things will be better then.

Friday, March 13, 2015

Safety bad for profits?

Have a read of this article.  In short, it reveals that insurance companies are concerned because they believe self-driving cars will reduce the number of motor accidents and thus adversely affect their bottom line. Also mentioned is auto parts makers with a similar concern.
The article also claims that full self-driving cars probably won't be on the market until at least 2030.

Please note that the following is only my opinion.

Humans love to drive their cars. Even though the car may become pretty much self driving and certainly have a safety awareness and crash prevention system that can override the driver input, I fully expect that many people will still want to "drive" their cars because it satisfies a human desire for power and control. 

Next we have the technical side of things: how close are we to a fully self-driving door-to-door autonomus car? We are not there yet, there are still some serious challenges to be overcome: mainly the ability to self-navigate in less-than-ideal conditions such as unfamiliar roads, rain, snow, at night and any combination of them - in short, computer vision and sensing. Much of the autonomous driving currently being talked about uses an already existing digital map of the roads that the system can refer to: if the computer needs to create its own on the fly the car will need to drive slowly.  In short, the technology is not there yet for full autonomy despite all the hype and sales pitches.
Maybe by 2030 it will be around but I'm not holding my breath: the computer industry will need some impressive new tech to boost system power and squish a lot more processing power into a smaller space to make it work, and that is not even mentioning the software required: remember, this needs to be small, cheap and energy effficient enough to fit in an average car and not price the car too high for people to buy. 

Finally, and by far most crucially, the issue of system faults and failures is the elephant in the room: I once wrote a short story where a misbehaving teenager hacks the autodrive system of a car to go for an illegal joyride (he is killed of course): so far we have computers flying commercial airliners but that is still far from the prospect of having cars drive people everywhere on a daily basis. The roads are already packed with cars and trucks and that will not decrease in a hurry, so what happens when those cars either have their selfdrive system fail, or (worse still) get attacked or hacked by malicious operators, be they human or otherwise?
Don't give me any bull about hackerproof systems, there is no hackerproof system and at this rate probably never will be: unless somewhere like Iceland takes over the world in a peaceful revolution, we will all continue to be watched by paranoid secret agencies and they will demand and get "backdoors" into every electronic gadget we have, which means that non-official hackers can and will find those same backdoors and bend them to their own criminal uses. (Hey there's a story or two in there already! - oh wait, who needs fiction? It's already really happenned!)

I'm guessing that by 2030 most of this won't matter anyway: other things will come that will make the whole issue trivial or at least secondary, for instance the endless increase in the price of oil because it really IS running out and only a few people with power and influence are doing anything about the boatload of **** coming down the turnpike towards us.

Keep fighting over those deckchairs on the Titanic, guys. You deserve the best view.  

Wednesday, March 11, 2015


No-one ever needs to mine coal, drill for oil or gas, dig up and refine uranium or anything else to burn or react: all the energy we could ever need is below our feet.

If people want to have electricity made in big installations that can run 24 hours a day seven days a week that do not emit any pollution, burn anything or create toxic or dangerous waste material there is only one answer: Geothermal power stations.

All of the technology needed to make this work almost anywhere on the Earth's surface has already been tried and proven.

All we need to do is drill a deep enough hole, pump water down it and steam will return which we can then use to generate electrical energy. The only “waste product” is heat, which every other type of power plant (apart from hydroelectric) creates too.

The drilling might cost something but the hole does not close up or “run out of heat” (unlike an oil well) and the water we sent down the well to heat up and drive the generators can be cooled in radiators and sent back down again so there is no need for a continuous supply of water or heat pollution of rivers. It is cheap in terms of generating systems and safer than any other power system (apart from maybe solar panels). All of the generator parts are available off the shelf.

So why isn't this happenning? Why aren't all of the countries of the world installing geothermal power plants? 

There is no technical reason why not.