Thursday, February 28, 2008

you know, patently speaking, if you started now, the Internet would never happen

so fighting patents (defending and attacking) is very like fighting a PhD (examining or
defending:)

There's a bunch of novelty required, and someone else has to be able to reproduce the results from the document (dissertation), and some of the intellectual tricks of the debate are similar - indeed, for people familiar with UK vivas (an almost uniquely sadistic approach to the game), one even sees people with documents full of yellow postits - the whole gladatorial style of experts in the dock being deposed (US terminology) or CXed (in UK) is also a bit like the Scandinavian defense where other peple than the actual examiner and candidate fight it out.

the process is simliar too, to an academic conference programme committee meeting - various people have reviwed a paper - not all are in the room - there are advocates of the paper and there are opponents = proponents (provendus in ductch PhD defenses) might argue novelty then an opponent will counter with insufficiency, or a proponent might argue sufficiency, and another oppoenent might argue prior art (unoriginal).

Here's an idea - why not HIRE programe committees to judge patents (see another posting on another blog on peer to patent) ? they have the skills and they have the time. and academics are so much cheaper than lawyers coz they love talking for free (my better half tells me som as do my kids).

Meanwhile, I quite like wikileak thing, but even more relevant is
http://trolltracker.blogspot.com/

Why? well people submit pointless papers to conferences (papers they know will be rejected) just to get feedback - for thoughts on this and related problems of checks and balances (and cheques and bank balances) see a forthcoming paper Keshav, I and Nick McKeown have been cooking up on aligning incentives in academic conferences - a lot of simular work is going on in trying to revise and reform patent law.

I can see a new conference (indeed I might apply for a patent on it)... ... ...

Wednesday, February 27, 2008

Gender difference in strategies for getting un-lost

so there's this old cliche about what men and women do differently when they get lost while driving in unfamiliar territory. Stereo-typically, women stop and ask someone, and men don't.

Here's a possible explanation based on a talk I saw at Ecole Polytechnique last week in Paris.

Hunters use a search strategy for finding prey (I suppose hunter gatherers do too for finding nts and berries and stuff), which entails alternating between local accurate surveillance of a small area, with a levy walk (random walk with path length eponentially distributed, not uniform). So in evolutionary terms, you might expect male humans to have such a strategy, whereas female humans might be in the village looking after kids (look, this isn't a proposal, so don't assume I believe in biological determinism as a prescription for social organisation:)

Anyhow, so assume a male car driver is running such a strategy. Stopping and asking the way would mess up where he had got to in the algorithm, and I can imagine there is a cognitive burden to searching like this, which would map into annoyance if interrupted.

If the population density had been higher when there was survival value in finding onesway, hen presumably the strategy of stopping and asking the way would perhaps have dominated both genders, but once people got to be "post-darwinian"
(i.e. where most people survive to breed irrespecive of "fitness"), then such selection doesn't happen, and we need to acquire better strategies through nurture...

An interesting question: Is there a sweet spot in population size where communities are more likely to be helpful to strangers finding their way, and larger, where alienation kicks in and people don't help? anecdotally, this seems reasonable to me...when lost in africa, I've had lots of help. when lost in NY, Paris, Munich, TOkyo, Sao Paulo, not a lot:)

Sunday, February 24, 2008

9 plots and Dunbar's number - computer science, anthropology and literature

It has been said that there are only nine different plots (lots of people claim seven, but I think this is just for mystical reasons), and after hearing and Robin Dunbar talk about the social structure of human societies, I think I can explain this.

Social groups organise in a hierarchy of trust of degree (3+epsilon), so you get a social group of size (3+e)^n (1, 3, 9, 27, 81, 243) roughly - the epsilon varies with smartness and species. Dunbar also observes that the neocortex size of primates directly correlates with effective maximum group size (chimps 40, humans 150), and that this shows up in human societies in may many organisations (villages, armies, companies that are succesful). The explanation is that this is the maximum size that a group can maintain complete cohesion over, since everyone in the group has a model of everyone else - this reflects a species ability to infer intentionality (humans can do 5 levels reasonably well) and therefore trust relationships. One aspect of this is that humans have spoken language which abstracts, and therefore accelerate (expand, in graph terms) the range of grooming (essentially, by being able to gossip) rather than relying on physical grooming as a mechanism to achieve pairwise (or indeed, (3+e)-wise) cohesion.

OK - so now consider how we encode (neurologically speaking?) a simple description of another person's behaviour? We can use the distributed systems folks' model- BAR (Byzantine, Altruistic, Rational) - this has a game theoretic interpretation - but is also (I think) respectable in anthropology terms -

- Rational is basically selfish (think, selfish gene) and is just what Nash said. selfish is not bad since it is predictable, but it isn't "nice" in the normal selnse - it is also how naive economists describe the "market" and claim it is good - bah humbug

- Altruism is what happens (often, not always) when an individual behaves for the "greater good" at a disproportionally negative cost to themselves.

- Byzantive is mad/bad/sad - bonkers, people that do random harm - in network terms, hackers, script kiddies, whatever, that don't do it simply for personal gain.

Interestingly enough, one can reasonable assume that anyone can choose any of these behaviours, but that various incentives keep one in a particular state (kinship keep you mostly rational, occasionally atruistic), and social pressure will keep most people rational, but, and here's an observation about cities and the internet and the p2p systems and online games and facebook, once a group size is too big, there are enough places for people to go to other social groups, that there are ways one can be a serial byzantine bad person, and still survive - alienation, disaffection, etc, we all know about this - at a certan scale, if the rate of such behaviour is high enough, one sees social collapse...

ok - so in literature, (plays books films) many stories have 3 main characters. A plot consists of social dynamics. Lets think "eternal triangle", or "three princes" or any other archetype (or if you like, shakespeare tragedy, or wuthering heights or lord of the rings).

So 3 characters, 3 states- change 1 state
(gollum helps or hinders, heathcliff loves or hates, othello trusts or doesn't trust)
gives 9 plots.

QED

indeed, classes of literature (jacobean tragedy - everyone starts luvvy, everyone dies) can be captured in this simple, elegant and ultimately futile and reductionist model:)

digital signatures and bots...

here is another really really bad idea that I really should patent, not.

I am forever being faxed documents which I then sign and fax back, thus wasting 2 copies (the blank input at the originator end and the signed input at my end)

what we need is a robot arm driven by a robot glove, and a fax protocol between the glove end and the robot end - then the document originator puts said document into the
faxsignator (TM), and I get a "ding" at my end, put my hand in glove, take up a pen, and "sign in the air", driving the bot at the originator end to reproduce my signature accurately on the original piece of paper.

Green and legal...surely?

Sunday, February 17, 2008

gerrymandering the news...

I just finished the Nick Davies work Flat Earth News - see also
http://www.flatearthnews.net/

quite a few of the arguments about PR are actually
previously extremely well analyzed in the book by
Jerry Mander (real name!) on TV
viz:
http://en.wikipedia.org/wiki/Four_Arguments_for_the_Elimination_of_Television

I was surprised this wasn't cited as it was written by an insider and
is very incisive...and around some time ago!

I thought the best material was the analysis of the construction of a
misinformed public awareness in US and UK on the buildup to war with iraq -
having just finished Alastair Campbell's fairly obviously carefully edited
diaries, one has to wonder at the brazenness of it all!

For me, what a lot of people working on traditional channels of information
vastly underrate is the accuracy, timeliness, and evidence of provenance
available on wikis and blogs and related content - for info, I recently put in a
proposal to the research councils in the UK to get funding to be a "media fellow"
(to improve public understanding of computer science) and proposed using
myspace/facebook + wiki/podcast- the reviewers all said "oh this is fun but i
should get a production company and do it on radio and TV 'properly'"!!
I was amazed - gven I was proposing something to reach "yoof" who watch/listen
to virtually no documentary programs on the legacy media, I thought my model was
relevant and plausible.

I guess the traditional media's continual assertion (without decent evidence) of the inaccuracy of information on the Net (as opposed to on TV, Radio, Newspaper) is because of their fear that with the right search tools and a little discernment, the average punter can find out a whole lot more about what is Really Going on via Cyber space, than via any of the PR, Security Agency and Corporate mindspeak that dominate the legacy media....

Alas...

Mr Davies writes quite readably, albeit a tad repetitive (I guess quite a lt of the material was edited from other sources - I'd ask for minor corrections if it was a PhD dissertation:)

Tuesday, February 12, 2008

primeval knowledge - when anomalies are commonplace...

so when you get an anomaly every day, surely it isn't anomalous anymore? eh, eh?

so what is with primeval anyhow? it looks more and more like a battle of boy and girl bands.

the "=" symbol should be read as "is easily confused with"
is a nice quote in John Day's new book on Communicatiosn Architectures (patterns thereof)

Saturday, February 09, 2008

founders at work - possibly re-imagining transformational government?

Founders at Work is a fantastic book with a huge array of internet talent on display, discussing the way their startups worked (and in some cases crashed and burned) in fascinating technical, social and economic detail. This is essential reading for anyone in the area - a few (amazon) reviewers commented that the book lacks analysis - it is true there's no pontification, but all the interviews are structured similarly, so that the analysis is emergent by seeing different gurus give their respective answers to similar questions.

For me, one crucial area is how these folks often didn't have insight into their own success (reasons for) until afterwards! e.g.

how many of the cool companies built 3 (or more) things before they
actually found the thing that made it big? often they didn't even
realize or understand what it was they had that made the thing succeed!

examples:
paypal's fraud detection, (not the e-cash)

hotmail's bottom of email banner viral advert/market, not the firewall busting

gmail too: auto-completion, not the search (wasn't even new - i was using
glimpse to search my mail and desktop/file systems for years before)

yeah, even google's click-thru advertising (not page-rank, which was only
marginally better than altavista at stoping people artificially boosting their site)

and, landsakes, RIM started by building realtime LANs before realising Mobile Email might be quite big:)

awesome:)

Tuesday, February 05, 2008

crossoverlay optimisation

new idea - combine cross layer optimisation and overlays - whaddya get -
crossoverleyisoptimality

Monday, February 04, 2008

a true history of wayback web browsing

so i live and work in about 6 places right now (2 labs in Paris, plus apartment, 1 London, 2 Cambridge) and typically am not in any one for more than a few days at a time. Flipside is that it is sometimes a few days between visits to any given place - so I tend to leave myself logged in (perhaps lock screen) on a Mac at each place - and so when I next show up, I have a bunch of tabs or windows open on activities that are a few days stale - e.g. news.bbc.co.uk, my facebook account my gmail account, a hotmail spaces and myspace - so on each of these, i see a snapshot as of 3,4,5 days ago - fascinating stuff - espeiclaly to se lcalization (sites dont show same thing - depends on your locale yur client ip address reverse looksup to...)

Friday, February 01, 2008

sexual rejection and journalism....and life

so andrew marr, in his fine book about the trade of journalism, says that being fired in a journalism job is like sexual rejection - it occurred to me to ask (since he said journalists are frequently fired) how often he had been rejected sexually. Then it occurred to me that this is actually quite an interesting question to ask anyone (especially any male one:) - how often do people "get it wrong" and chat up someone, only to be told no? I imagine some people are really accurate at picking people for whom attraction is mutual, and some might be really bad....and are men and women different in the average accuracy of their guesses.

while reading this book, the chap next to me on eurostar was reading a manuscript for a solo violin piece, and reading a book about it too - very distracting - it occurred to me that reading music is very like dancing about architecture, or watching star trek movies on french TV with Klingon translated into french - furthermore, someone was talking to someone on their cell phone in french, and i swear the conversation was one of those "so I says to him, i says, ...then he says to me, then she said, oh you never, you aint never gonna believe this..." blindiingly awful...

luckily, i was able to finish my book, read my email and get some fine beer

Blog Archive

About Me

My photo
misery me, there is a floccipaucinihilipilification (*) of chronsynclastic infundibuli in these parts and I must therefore refer you to frank zappa instead, and go home