Saturday, August 30, 2014


First, let me assure you that I am not a fan of "Machine intelligence": I am a skeptic about all such ideas because I have never yet seen any system that even vaguely resembles human awareness of its world or human mental ability.
That said, there are things computers could do that might improve lives if they are permitted (or, cynically, if someone pays for it to be done) . . and this is the sort of thing IBM is now trying out in this article.
This could be good, dealing with the mountain of stuff churned out by people in labcoats, boiling it down automatically and deriving or "mining" information from the mountains of data . . . .
but something sticks in my mind about this.

I recall an old scifi short story where someone invented the "de-legaliser" - software that read legal documents and rephrased it in simple terms that anyone could understand. Of course, the next step was to make the software work in reverse . . . . and by then the machines had started taking over the whole legal business.

I am curious to see how far this whole circuit circus goes.
Just remember that there is no actual research going on here folks, and also note that not all research papers are true and accurate:
Is the software smart enough to know when it "reads" faked research? 
How does it deal with the minefield of legal ownership, copyrights and patents?

One can imagine a scientist who has just completed his life's work only to be trumped effortlessly by a machine in a few minutes.
Hmm. . .


Friday, August 15, 2014

Science or Fiction?

All of us Space nuts know that what you really need to travel in space is a form of space drive that does not involve throwing matter out the back of your ship as rockets do. First, it is limited: once you run out of stuff to throw away, that's it.  Second, it means you need to keep lots of explosive, dangerous stuff in your ship while your drive is operating and as the stuff gets used up it changes the ship's performance.  Oh, and it's very limited.  If only we could just use energy itself to drive our ship through space we could equip it with generators, batteries and solar panels and just Go. No need to stop and refuel anywhere, no huge tanks of dangerous liquids required, just thrust at one gee aslong as you like and you can enjoy earth-normal gravity while getting there much, much faster than any rocket ship. 
Well the idea has been around for a long time, but recently people have claimed success doing just that: making a reactionless, exhaustless thruster.
Have a look here, and here courtesy WIRED.  To date, the total thrust output of the test devices is tiny, but even so, it already beats existing satellite systems and it never needs refuelling.  Then consider  the potential of a more advanced model using superconductors, which would work very well in space since the main requirement of superconductors as they are today is near- zero temperatures, which are already plentiful (as it were) in space.Provided you can take enough electric power with you, you can get anywhere in the solar system fast enough (and with artificial gravity, esssential for healthy humans) to do some proper exploring. Okay, so I am going out on alimb here postulating a reactionless thruster that can push one gee, but even with a lot lessthan that we could at least make a start. 

Then there is the backlash: according to some people this drive cannot work "because it disobeys Newton's laws of motion": Hilarious!  I don't care a damn what critics say: the proof is in the results and so far I can see that  there is a principle here that might work - and there are at least two science labs working on it so I think it will come to pass as real.  Actually, the very fact it does work will once again shake up the world of science and force people to question their blind assumptions.  Great!
Actually the "how" of it does not matter: only that it does. I recently read about big arguments by scieintific types about how ariplanes actually fly: the classical definition of how a wing profile generates lift is  not just open to argument, it cannot work that way: yet every day thousands of aricraft certainly do fly so it really does not matter so much - the truth seems to be a good deal more complex than the simple explanation I grew up with . . . . .

Saturday, August 9, 2014


You have been deceived: machines and robots will not "take your jobs", they are not capable of that.  Humans running Corporations and Governments are doing that. 

Once again, I read the same old fears and expectations about robots:
How they will take our jobs
How they will soon be so smart they will replace people
Where is the sense from these people?

Manufacturing truth and illusion

Myth No.1: Robots are about to take over production and will soon replace everyone in manufacturing.

It seems that some people believe what I call “The George Jetson Idea Of Automation”: in the cartoon, George goes to work every day and presses one button that starts the machinery of his employer. That's it, nothing more to do. I have followed automation for many years and have worked on automated production lines and machinery and none of them were ever like the George Jetson's world – or will be in the forseeable future.
First and most importantly, the machines are just not smart enough to work by themselves and at the rate of computer progress we see today they will not be smart enough to diagnose their own faults and correct them for a very, very long time.
Don't be fooled by application specific devices either: sure, Google can get a car to drive itself but that is after many humans spent millions of manhours developing it, and as of writing this, it is still in development – and this is application specific : you could not take that same computer driver and then throw it in an office chair and expect it to read and write reports. Computers are really, really dumb. They can only deal with things in their area of experience.
Every so-called automated machine needs human oversight. Not “should have”, must have: machinery and robots are not cheap and when they go wrong, they really mess things up.
There are ways to cut the number of human overseers and operators but these result in some very negative outcomes for the factory: first, you can set up your machines to only produce the same product over a long period – but you still need to check the output and make sure the machine is doing what it should, and even then, a company that produces exactly the same product year after year is unlikely to compete against other companies that constantly change theirs even if the changes are only cosmetic because humans like variety.

The very idea of manufactured items all coming off a production line all uniform and perfectly identical is a myth. Complex machines don't ever run smoothly for very long before they need human attention, often because the complexity means there are so many things that can vary but also because things shake, bend and wear and so the first fifty, hundred or thousand items might be fine but gradually the output will get worse and worse with no human overseer to stop the line and fix it.

What has really happened in manufacturing is that corporations that have no moral or national limitation have moved their production to wherever is cheapest for them, using the cheapest human labour possible to get the highest profit margin possible. 

Myth No.2: Robots will soon be smart enough to replace humans in technical and engineering jobs.

Let's take a serious look at this: the claim is that computers will be “smart” enough to engineer products and processes and develop new products, their manufacturing lines and logistic and supply systems that will get those products to the end users.
Who writes this stuff? No computer has the capacity to do any of these things now and I don't see any in the near future either. Consider the matter of “artificial intelligence”. We don't know what intelligence is, so lets ignore that for now: consider the processing capacity of the human brain.
One estimate of note is that in 2002, if every computer in the world were connected together running the same master program, they would equal the processing power of one human brain. One. We don't know what that program would be, but we can be certain that even if we had a computer on our desk that had the processing power of a human brain (and the estimates of that vary too) it would be no use without the “program” that would make it “intelligent".

The companies making computer chips are not making endless progress toward faster and more complex computer chips because the nature of matter and energy create limits that are not easily avoided. Already they are looking to change the whole architecture of computer chips to get around the limitations of existing chip production. 
There is no “Moore's Law”: it is all marketing.

What we are is the result years of learning in a mobile body with sensory information coming in continuously and being handled in ways we don't understand which we then use to generate new actions and reactions. There may well be information and/or processing methods that are built in from conception but the nature of them is open to debate.

A second, and perhaps more important problem for robot makers of the future is memory capacity:
human memory is nothing like computer memory – the memory system of a human brain is mixed in with the thinking system, there is no discernible division of function unlike a computer. In computers, we have storage areas that are completely discrete from the processing area. Then there is memory reliability. In current computer systems, up to a quarter of the whole storage capacity of your computer is used to correct errant data and fix errors when they occur when copying or transferring data from one place to another. Human memory is far from perfect as any court case will remind you, but the method of storage is different.


You have been deceived: machines and robots will not "take your jobs", they are not capable of that.  Humans running Corporations and governments do that. 
The real issue is what we humans will be doing in future for jobs and money.
Machines are very good at doing boring, repetitious and dangerous jobs for long hours - but they still need human oversight. . . . The REAL issue here is how we, the common and not-so-rich people, can survive in the future if sociopathic people and corporations have swallowed up our sources of income, work and contribution in their insane machinery of acquisition.

People need jobs. I know you might say “We don't need jobs, we just need money” - but in my experience, people need to contribute to society in some meaningful way, and recieve from it return. Those of us who have a sense of fairness and morality know that: even if it is trivial, actions that affirm our membership and right to contribute to, and receive goods and services from society keep us connected and sane.
Prevent people from participation in society and there won't be any society - or they will make their own new one.

Big business does not have the right to drain our society of income just so that we can all (well, those who still have money) buy cheaper appliances or furnishings.
Already small people have begun to create their own self-support groups, bypassing the Big Business funding and operation models that have cut them out of the services. I expect this to continue and expand - the test case is currency. If new electronic currency develops like it should, the Megabanks will be cut out of all the billions of small trades going on since they demand a fee for each trade – and by using blockchain currency e.g. bitcoin, they are not needed or wanted.
This could result in a two-level economic system where only the biggest trades will involve the old money while most of us use purely electronic currency for day-to-day trading. This creates a lot of questions, but for now I will only say that something must change, the only question is wether the change will be peaceful or bloody.