Libre Planet 2009

Audio recordings from the Libre Planet 2009 conference have been made available. The Jeremy Allison talk is quite fun.

Software engineers at work

I came across this curious university department, called the Software Engineering Observatory.

"researchers can investigate how software developers work in teams, deal with industrial clients and handle the plethora of problems that arise in group projects with tight time-scales"

I expect that an anthropological study of software engineers in their natural environment won't reveal very much of interest, because software itself is metaphysical and the tools required to create it consist primarily of...a lot of thinking. Here's what I imagine that an observer would see if they were to do a time-and-motion study of me.

• Stares at laptop screen
• Taps the same key a few times
• Two hours later, still staring at laptop screen
• Drinks coffee
• Silent, staring out of the window
• Seems to be doing nothing
• Looks like he's praying
• Reading email
  
• Doodles some wavy lines
  
• Staring at laptop screen again
• Hasn't said a single word all morning
• Deletes some code
• Strange patterns appearing on laptop screen - they seem incomprehensible
• Subversion commit (deletions)

Of course sometimes I do actually write code, but probably at least 90% of the time I'm not really doing anything which an observer could directly see or record, other than thinking about the problem and how to solve it. Mechanics and fitters in the past often described me as either "sleeping" or "praying". The remaining 10% of the time usually is conspicuously industrious though.

Even in the most constrained, highly specified environments - such as an industrial production line - the process of developing software is always a creative one comparable to writing a novel, albeit in a very formalised language. Fundamentally it's about engineers creating and implementing solutions to problems.

On a related note I did once read a paper describing the typical eye movements (saccade targets) of users as they read through Google search results. It would be interesting to apply the same measurement techniques to someone writing software, although I think it would still remain difficult to infer from this exactly what is going on at the cognitive level.

Biota podcast review

An amusing iTunes review of the Biota podcast has been brought to my attention, as follows.

"This is the single most infuriating podcast I've ever encountered. I emerged from literally hours of listening not one whit wiser as to what artificial life really is, its level of development in the world or its future. Hours upon hours of the program are taken up with club business --- the need for another convention, improving the website, the best time to schedule a conference call across time zones. The signal-noise ratio was so low that I began to fear the entire field of artificial life was a bogus (sic) as Korzybski's general semantics of yore. I strongly urge Biota's host-producers to give a listen to podcasts like Talking Robots and learn what a truly informative program is capable of."

There's probably some truth in this, although it is perhaps being a little unfair because there are at least occasionally interviews with people who talk about their particular Alife developments.

Emergent grids

Grid cells which constitute part of the localization system of the GROK2 robot are formed from a pair of theta waves (3-6Hz), whose interference produces a moiré grid. Grid spacing is controlled by small changes in theta wave frequency and angular offset. The grid dynamically rescales as the robot's speed is altered, due to variations in one of the theta field's frequency.

On the grape vine

It looks as if automated pruning of grape vines is now possible, at least under some fairly controlled conditions. Interestingly this uses stereo vision rather than lasers. As a stereo vision application goes, being able to detect objects against the background at short range is well within the scope of current technology. You can see the cameras attached underneath the forearm, and I guess for this type of job a scanning laser would be just too big and clumsy.

It is nice to see the kinds of robotics which previously only occurred inside factories gradually making its way out into the wider world, and as robot arms get cheaper maybe other varieties of agricultural or horticultural labour might come within the scope of automation in the near future.

"You have run out of our money"

A very apt speech for the times. Most people who are paying attention, and who aren't held in the mesmerizing trance of mainstream media nonsense, are already aware of all of this stuff - such as that you can't borrow your way out of debt and that the current policies are basically selling out future generations.

I received my council tax bill recently, and surprisingly it has only risen by around 1.5%. In the last decade I can't think of any year where the tax increase has been less than the official rate of inflation (currently 3.2%), so I'm assuming that this is the calm before the taxation storm begins.

Open seeds, open source

Complex patterns from simple origins

One regular grid is rotated over another, and the resulting interference pattern, or moiré grid.



Patterns like this may actually form in the brain as the activity of neurons become phase locked to two or more interfering oscillator sources. It would be fun to try this in 3D, for example, or 4D if the oscillators vary dynamically in frequency. I recently read an interesting paper on grid cells and a possible computational model of how they form as an animal moves through its environment. Such mechanisms might even have some application in the robotics development which I'm doing.

Is Facebook the new Stalinism?

Facebook may be all the rage at the moment, second only to Twitter, but if you have a Facebook account it's a good idea to be careful about what information you make public. Ever since Facebook was opened up beyond its initial student user base there have been rumors that sinister forces may lurk behind the seemingly friendly facade, in the form of US government spooks. Whether there is any truth in the rumors or not it seems likely to me that other organizations, especially insurance, recruitment and marketing companies, may be interested in what happens on Facebook and may be given (or already have) access to your data without your knowledge or permission. Hence I think it's a good idea not to mention health problems or other events which might be judged by insurance companies as increasing risk, or might be used by employers or recruitment agencies to disqualify you from being considered for a job. Although Facebook may appear to be just a cosy circle of friends - the electronic equivalent to chatting in a pub - it's really much more than that, and potentially also has a dark side.

So, it comes as no great surprise that the UK government "is considering" monitoring Facebook and other social networking sites. In this context it's safe to say that "is considering" means "will soon be", just as the government was once considering monitoring web sites, or considering monitoring email. You might think that this is fairly harmless, and even believe the notion that the government is only interested in catching criminals or terrorists, but if you're unlucky enough to have been recently made redundant and be claiming some sort of state benefit remember that this would give the social security beaurocrat behind the desk full access to the details of your life, including who your friends are and what you've recently been saying to them. This level of access to the minutiae of people's lives would have been Stalin or Honecker's wet dream, because with knowledge comes control and power.

The main problem, as I see it, is that Facebook is privately owned with centralised servers. Where you have centralisation you also have a weak system which is easily compromised by sinister forces. What's needed is something like a peer-to-peer version of Facebook, where the data is transmitted in encrypted form and where users have full ownership and control over their own content. Such a system would be much harder for any central authority to snoop on, and you wouldn't have all the complaints about the layout changes.

The Tata Nano - I want one!

When it comes to cars I'm definitely not in the Clarkson school. I absolutely don't buy into any of the hype or the branding nonsense which many of my engineering contemporaries seem to swallow with tremendous enthusiasm. For me a car is a purely utilitarian machine for moving from one location to another.

The great thing about this car is its fuel economy and also the price. At a time when many people's wages are either being frozen or actually falling I expect that companies selling cheaper cars are going to have a distinct advantage in the market within the next few years.

The end of the independent music store

A rather melancholy documentary about the decline of an independent record store. I'm of a similar generation to the guys in this video, so I remember the heyday of small record stores in the 1980s and 90s. The decline of this sort of business model in the early 2000s was a phenomena which spanned worldwide, coinciding with the rise of broadband internet and the iPod. Once I saw and used Napster in 1999 it seemed clear to me that the conventional record store would soon no longer exist. Even at that time it seemed reasonable to believe that internet download speeds would increase and unmetered ISP subscriptions were beginning to appear making internet use far cheaper than it had been previously.

In the late 1990s I remember that there were a few small record shops in York where I was living at the time, but by 2005 most of them had closed. The big stores, such as HMV, have survived mainly because they have diversified particularly into computer games but even for these goliaths it looks as if the days of browsing through isles of CDs will soon be numbered.



It's sad for the old timers, but actually we're living in a golden age of easy on-demand music availability. They talk about the demise of a record company "developing the artist", but if artists sell to the public in a much more direct way, without the multiple layers of middle men, they're free to develop in any direction they choose and may also have the advantage of actually owning the copyright on their own work (which typically didn't happen in the olden days). Music now is more democratic, and less under the stranglehold of a few big media companies. As far as music is concerned we're living in a post-scarcity economy.

Linux foundation videos

There appear to be a whole flotilla of videos recently put up by the Linux Foundation, which are more or less promoting Linux. These range from the geeky to the ultra geeky to the just plain painful to watch. This one was vaguely ok, but it's tempting fate to say "there are no problems".

If the goal is to promote use of GNU/Linux over other operating systems the result here seems to be a big FAIL. I'm not a marketing person, but when devising some promotional media you need to think about:

• Who is the target audience
• What are the main advantages of Linux over other OSs
• What are the main problems faced by non-Linux users
• What are the core values surrounding Linux and how might these appeal to the target audience
  
• Is this video likely to lead a non-Linux user to try using Linux for the first time

If I had never used Linux previously, but only vaguely heard about it, I don't think any of these videos would even remotely encourage me to try it.

Probably the best way to promote Linux use would be to show someone using Windows, battling with spyware/crapware/adware/badware/lameware/disableware/malware/viruses/rootkits, then trying to download and install free "registry cleaner", anti spyware or anti virus programs and getting into even more problems and their computer slowing down to a crawl with numerous popup warnings. Then they try to re-install Windows, but they can't find the CD because the computer didn't come with one, then borrowing a CD from a friend, then spending hours trying to reinstall (picture of a clock with the hands spinning round) and trying to find license serial numbers.

Then after all this stuff has been depicted there's a voice-over which says:

STOP!

Is your computer wasting your time? (nods)

Want more control over what gets installed and what doesn't? (nods)

Had enough of dodgy software and stupid serial numbers? (nods)

Fed up with "security" software holding you to ransom and slowing everything down? (nods)

Then you should become a LINUX USER.

(show someone using Linux)

In big letters overlayed on the screen:

NO VIRUSES

NO ADWARE

NO REGISTRY CLEANERS (no registry!)

NO NONSENSE

The thinking machine

An amusing documentary about the history of AI, with some nice footage of various vintage systems and commentary by their inventors.

The doors of perception

Detecting a door handle and opening a door.



In terms of mechanics and electronics there's probably nothing going on here which couldn't have been done several decades ago. What makes this possible is the fidelity of the robot's spatial awareness, using its scanning laser rangefinder - a technology which has really only matured to the point of being useful within the last ten years. As robot perception improves so too will the range of possible jobs which can be automated.

There's still considerable doubt in my mind over whether laser rangefinders will be deemed safe for use in domestic situations. The low power lasers appear to rely upon the human blink response as protection, which seems somewhat questionable, and whether or not it would also apply to pets such as cats and dogs also seems unclear.

Programming languages

It's always a good idea to keep an eye on which programming languages are becoming popular, and check out anything which might be worth learning or merely fooling around with as entertainment. I found this site which has quite a comprehensive set of graphs of programming language statistics.

It looks like C, C++ and Java are top of the pops. I expected C++ and Java to be up there, but I'm surprised that ordinary C remains so popular, in spite of at least two decades of relentless bombardment with object orientated dogma.

Soup of the day

Robot fetches stuff

This looks like pure dead reckoning after initially registering with a recharging point, so I bet there were many out-takes. With added vision the behavior could be made more robust. Also the fridge door gets left open!

We also really want to see a cocktail version of this.

Batteries that "charge in seconds"

If this turns out to be true, then it could have many beneficial applications, including making electric vehicles much more practical. Certainly, improvements in battery technology will also be a critical enabling factor for robotics. Like many computer users I'd also like to have a laptop whose battery lasts all day on a single charge, which would enable a genuinely wireless lifestyle.

However, it's worth remaining skeptical about technology which is still at the research stage. Most promising sounding research breakthroughs never make it to market, either because they suffered from unanticipated problems, turned out to be too expensive to manufacture, or because they were deliberately blocked by companies who bought the patents and then sat on them.

Worthless money

Listening to the news on the radio today the descriptions of how the Bank of England has commenced its "quantitative easing" were quite surreal. Reporters were talking about it in an almost cheery manner, with absolutely no historical references or even the slightest hint that this might be anything less than totally financially responsible. Not once was it mentioned that this could be a dangerous/wreckless policy, and not a single commentator said "hold on, could this be a huge mistake?".

Searching around for references to "quantitative easing" I located this Ron Paul clip, which although nothing to do with the UK economy seems very apt if you replace the word "dollar" with "pound". I agree that you can't just create money out of nowhere at the stroke of a magic wand and call it "capital", or if you do you're on a rapid downward slide into third world economics and all the social chaos and impoverishment which goes along with that. Injecting large quantities of worthless money into the economy probably isn't going to improve anything, and stands a good chance of making the situation a lot worse.

Cocktailrobotik

Locating a cup and pouring a drink.



Described in more detail in this paper. Obviously there are limitations to this approach, but I think this is moving in the right direction. To begin with the robot has some foreknowledge of the environment, in terms of 3D models of different types of container. This might not be too impractical, since common kitchen items tend to come in a limited number of 3D shapes (cups, plates, cutlery, etc). Also, this isn't solving the general object recognition problem, but instead is more reminiscent of active vision, where the visual recognition is part of a larger closed control loop which includes the motions and pose estimate of the robot itself. In reality I think biological recognition of objects works something like this, in a very context sensitive and closed loop way with top town expectations being continuously synchronized with bottom up observations.

Recession and mental health

Just repositioning the tin foil hat for a moment, I wonder whether the "mental health problems" described in this BBC web site article will include people who are angrily protesting that the government/bank has stolen their savings or rendered their wages/benefits/pensions worthless.

Being gainfully employed certainly has mental health benefits, and you are at greater risk of psychological depression if you become unemployed. In the early 1990s when I was a fresh faced school leaver I remember meeting people who had been made redundant from jobs which they had been doing for a very long time, and which they had assumed that they would be doing for the rest of their life. These people seemed to me to be suffering from symptoms similar to post traumatic stress disorder, or even bereavement, in that they had been abruptly dislocated from their usual social context and placed into a financially austere situation where their previously learned skills and habits were no longer adaptive (or actually maladaptive). The lesson for me from meeting people like this was that it's not a good idea to strongly associate your personal identity with your job, and as far as possible to maintain interests and pursuits unrelated to work so that you have a more rounded personality and can survive such dislocations with minimal psychological disruption.

Brain emulation criticised

Tyler's main criticisms are:

Evolution designs messes

It's true that evolved systems are not easy for us to understand. Although the brain does appear to be modular to some extent, it's not the same sort of modularity which you would find in a piece of computer software or other conventionally engineered items. This makes simulation a challenge, but provided that the rules of adaptation are known not an impossible one.

Brain emulation would be technically harder to achieve than a more traditional engineering approach.

At present since we don't fully understand the requirements, either for a non-human artificial general intelligence, or for human brain emulation it's difficult to say which approach will turn out to be the harder one. The broad outline for brain emulation described in the video remains just that, and we don't know which aspects of neurons are essential to their functionality and which aren't. Neural models exist, of varying degrees of complexity, but whether these adequately capture the essence of what neurons are doing - i.e. that they are truly functionally equivalent devices - is not at all clear.

If it turns out that the functional essence of neurons can be encapsulated in a fairly straightforward way, for example using something like the Izhikevich model, then it could be that WBE is much easier and more straightforward to achieve than via classical engineering.

The economic case for WBE is unclear. Companies want systems with transhuman abilities, since they can already easily employ human level intelligences.

This seems like a fairly weak argument. If WBE were technically possible today I could easily imagine commercial uses of it, provided that the costs of running the computer systems are manageable. Human intelligence is expensive to employ, and is only partly an online system requiring significant down time for maintenance and consolidation of memories. A large part of the costs of running many businesses is human labour - increasingly cognitive labour rather than physical. If I could economically run several human brain emulations I could start a software company using the uploads (which might just be digital duplicates of myself) as programming labour. It's likely that such a company would have lower costs and be significantly more productive than an equivalent company employing biological humans.

This scenario isn't ideal though, since even emulated human brains would still suffer from well known scalability issues. Simply running large numbers of copies of brain uploads would not necessarily be very different from running a large corporation, with all the associated inefficiencies. This is because the human brain is fundamentally oriented towards a small tribal scale of existence. Going beyond these limits would require redesigning the brain simulations to some extent, such that they would no longer be strictly synthetic reproductions of biological systems.

Uploaded persons would be socially unacceptable, and subject to discrimination/enslavement.

Although probably true, this is not a strong objection to WBE in my opinion. I think biological humans would soon adjust to the idea of co-existing with non biologicals. It might take some time, and there may be trouble, but some form of eventual co-existence might result.

In the case of tech companies like IBM, brain emulation is largely a public relations exercise.

Even though some good work does appear to be going on in this area, I'd agree that from IBM's point of view it probably is more about marketing the company and attracting the most talented employees in a competitive job market. So far commercial spinoff applications from brain emulation research seem to be very few and far between.

Humans are crazy, with volatile emotions. Copying the human brain would not produce systems which are verifiably safe.

Once WBE becomes technically possible I think that aspects of human psychology which at present are poorly characterized will become far less so, and that it will be possible to engineer brain simulations which under most circumstances could be deemed to be "safe". If people's brains can be easily scanned and turned into a simulation it will be at least in principle straightforward to compare brains and find correlations with well known psychological disorders. Even if these disorders cannot be directly engineered out from the simulation without losing essential functionality, it may be possible to devise watchdog systems which generate alarms or modulate the operation of the system.

Since the human brain is the result of a long evolutionary process it is a more constrained system than a non-biological AGI. The space of all possible non-biological AGIs is much wider than the space of possible human-like brains or human brain derivatives, therefore safety risks from systems not directly based upon biology may be considerably higher.

The power of speech

Here's a demo of a robot using a simple speech interface. I did something similar to this years ago using the Rodney robot, and short stereotyped phrases can work quite well. It's when you start trying to use longer more complicated sentences consisting of more than a few words that things tend to come unstuck.

This looks like a fairly conventional trash can robot using ultrasonics for navigation. Ultrasonics will work ok as the primary sensors for SLAM, but the robot's resolution of spatial perception will be very low compared to a vision based system, and consequently won't be good for object recognition.

Bank of England "creates" £75 billion out of thin air

Amidst more euphemisms the Bank of England seems to be rapidly becoming the Bank of Zimbabwe. Here are the euphemisms which I've heard so far, used in the mainstream media (I'm keeping a list).

• Expanding the money supply
• Enhancing the money supply
• Quantitative easing
• Boosting the money supply

As an ordinary citizen if I "created" £75 billion using nothing other than a printing press and a lot of paper in a back room, and then started spending it as fast as possible "to enhance the economy", I'd be called a fraudster and I'd expect to go to jail - and deservedly so. It seems fairly clear that the media are doing their best with obfuscatory language to try to prevent people from realizing what this means in practical terms. "Creating" money in this way basically devalues the currency. It makes your money worth less, and is equivalent to directly stealing cash out of your wages or from your bank account.

"The main cause of hyperinflation is a massive and rapid increase in the amount of money, which is not supported by growth in the output of goods and services." - Wikipedia

Clearly this has to stop, otherwise the currency, and consequently the country will be absolutely ruined. If people's savings disappear in a sudden shock of hyperinflation, at a time when those who are out of work will be critically relying upon their savings for basic sustenance, there is going to be a lot of confusion, fear and anger - which even Gordon Brown won't be able to keep under control. Historically when such things occur people believe that mainstream politics has failed, and they turn to extremists who promise easy solutions, and easy scapegoats.

ESR probes FAT (and other dubious patents)

Eric Raymond examines Microsoft's patents in the TomTom case, and concludes that they're no real threat to Linux. It looks like the FAT claims will fail due to prior art, but even if by some improbable act of divine intervention Microsoft manages to uphold those patents it's completely trivial just to use a different file system on an embedded Linux device. As far as vehicle navigation systems go, the user absolutely doesn't care, nor do they really need to know what file system is being used. FAT is so archaic anyway that using a different system would be an upgrade in terms of reliability.

Microsoft's view of 2019

<a href="http://video.msn.com/?mkt=en-GB&playlist=videoByUuids:uuids:a517b260-bb6b-48b9-87ac-8e2743a28ec5&showPlaylist=true&from=shared" target="_new" title="Future Vision Montage">Video: Future Vision Montage</a>

This is indeed a dystopian vision, because it suggests that the world ten years from now will be exactly the same as today, only with slightly larger screens and slightly slicker graphics. Touch screens, mobile phones, eBook readers and PDAs are all legacy technology now, so this isn't anything new.

In this view of the future the robotics revolution never happened. There are no telerobots, and presumably the retired baby boomers, unassisted by mechanization, were abandoned to suffer the same fate as previous generations of elders. Hilariously, and somewhat sadly, people still manually push supermarket trolleys around and need a mobile phone to tell them in which isle the cheese & onion crisps are located. The job of re-stocking shelves also remains a totally labour intensive and expensive waste of scarce human resources.

As roboticists this ought to be a rallying call. Surely we can do better than this!

Also, a world in which Microsoft reigns supreme is probably one which is totally inscrutable and locked down by a no mans land of digital restrictions. Where people don't own their software, and have little real control or influence over what it does, or who gets access to their data. It's a world of closed, proprietary formats and user lock in - of digital disempowerment. There are no personal computers anymore, only corporate cloud computers.