Here is an audio interview courtesy of Marjan Grilj (Slovenian transhumanist association) that I gave in Milan on brain emulation and whole bunch of other subjects.
Currently a bit too busy with running around in Westminster as part of the Royal Society MP-scientist pairing scheme to blog much.
Accelerating Future » Toby Ord on BBC for Giving What We Can - a post on my saintly colleague Toby's very worthy project Giving What We Can.
"Toby is participating in what I glibly call “utility war” — a worldwide war not for money or power, but to achieve the greatest good for the greatest number (positive utility). This could be the war to end all wars. A war we can be pleased to fight."
However, what we really should try for is a utility arms race. The more people are efficiently giving (I expect most charity to be driven by social signalling rather than a rational desire to improve things) the better things will likely become. Note that Toby's very utilitarian approach doesn't hinge on getting everybody to go along with it: the more the better, but every rational donation is on the plus side. The trick is likely to hi-jack the signalling by making it possible to signal strongly and honestly through the charity, but ensuring that the donations actually go to efficient goals.
Dominic uses Toby as a nice example of that at least some ethicists are ethical. And I have been deeply impressed by the enormous range of cost-effectiveness of charities - if it is true there, it might be true in a lot of other social domains too. There could be enormous improvements just waiting to be found.
Sergei Maslov writes a viewpoint Power laws in chess about the paper
An interesting point made by Maslov is that the branching ratios of the game tree seems to approach the distribution 2/(pi sqrt(1-r^2)), which has a peak close to 1 - most positions have very few likely subsequent positions. This may be due to the database used mainly contains games by skilful players, who do their best to move the game into situations where the opponent has few choices. So if he is right, here is a signature of intelligence in the game tree statistics.
Apropos power laws, I just read Power-law distributions in empirical data by Aaron Clauset, Cosma Rohilla Shalizi & M. E. J. Newman (SIAM Review 51, 661-703 (2009)). A nice extension of their previous very useful paper about MLE estimation of power laws, and a stern reminder to use careful statistical methods before claiming power law properties. I suspect Blasius and Tönjes would do well checking their data with maximum likeliehood rather than logarithmic bins before claiming as they do, "Stretching over 6 orders of magnitude, the here-reported distributions are among the most precise examples for power laws known today in social data sets."
I gave some input to a report written by the Future Laboratory for Oracle about "Capitalizing on the digital age". Not very futuristic by my standards, but still full of interesting small ideas that make sense of things.
For example, the trend towards polarization of quality seems real - lots of free, participatory, unpolished and unedited stuff that people like, but also a higher premium on the really good information. I cannot live without both Wikipedia and The Economist.
An interesting aspect is that privacy is becoming tradeable. "Consumers to Swap Privacy for Customization" as one article put it. Privacy is not just (potentially) a right, but something we negotiate, trade and use. This of course makes it even more important for companies and governments to retain the trust of their users. People are voting with their (virtual) feet much more today, and they are better at telling their friends who broke their trust.
I feel moved when I see the statue of Hierta,
but only smile when I see the parade of kings.
I feel proud when I see the banknote,
not over where it was printed
but because I am the same species
as the gardener monkey.
I love well-executed crossovers between genres, arts, species, whatever. Finding fruitful or bizarre analogies and synergies in disparate domains is so fun! Here is one of the most amusing crossovers I have seen so far: chess-playing novels.
The authors encode novels as sequences of 2-tuples of characters, and map these tuples onto squares in a spiral pattern based on their frequency. This way the most common tuple corresponds to a center square, with more rare tuples near the edges. The program searches through the text for the next tuple that corresponds to a square with a piece that the player can move, and then finds the next tuple corresponding to a valid move. This ways novels can play chess against each other.
In general the novels are not great players, but they are not random monkeys banging on the keyboard either. They claimed French novels to be tougher, individually and as a team:
In our last “world cup,” a five-novel French team (led by Alexandre Dumas) trounced the competition, defeating the second-place English team by a 20% margin (the UK was hurt by poor performance from Disraeli’s The Infernal Marriage of 1834), and leaving the also-ran Italians looking rather like novices.
As to why Constant’s thinly veiled exposé of the intricate erotic politics surrounding Madame de Staël should prove so formidable an adversary, that is a perfect mystery (though it should be noted that both Frankenstein, in the second edition of 1831, and Goethe’s epochal 1774 Die Leiden des Jungen Werther, playing black, successfully fight off forceful opening gambits and record victories against this opponent).
They end with a truly fascinating idea, something for superintelligent Oulipians to try their hand at:
In closing, it is perhaps worth addressing the possibility of composing novels specifically calibrated to win chess tournaments convened by our application. The prospect of a “grand-master,” a kind of all-purpose novel-killing novel, while alluring, strikes us as beyond reach—indeed as probably a formal impossibility. But a novel written to defeat some other specific novel would appear to be an attainable objective, though a little thought suggests this would be by no means a trivial undertaking.
I found the talk very enjoyable, and it was a bit unsettling to learn that there are AIXI-versions up and running today. Yes, they are weak and merely playing pac-man, but it was like hearing that somebody had actually made tiny, safe black holes in their lab. It was also exciting to hear just how much has been happening in the neuroscience/neural network world while I have been distracted.
Shane's Halloween scenario was fun, because it is the closest thing I have seen to an AI roadmap. And it is quick. Of course, AI is *always* 10 years in the future :-)
Practical Ethics: Speaking truth to power - about the affair caused by the sacking of professor David Nutt from the Advisory Council on the Misuse of Drugs since he had criticised how the government has been systematically ignored or distorted the scientific evidence on drug harms.
It is an interesting demonstration of how the political sphere is less interested in truth and evidence and more in maintaining the social and political structure. "Independent" advisers should ideally provide trustworthiness but not challenge government policies, and should definitely stay loyal to whoever appointed them. At the same time scientists may be less able to do politics well: it requires skills of negotiation and compromise. Scientific results are not open to negotiation or compromise (at worst, they are open to interpretation - but then they are not very good results).
Nutt has written quite a few interesting papers and made fun points.
He has considered whether we could re-engineer alcohol to retain the good effects while limiting the bad ("Alcohol alternatives – a goal for psychopharmacology?" Journal of Psychopharmacology 2006 20: 318-320 - see also the discussion in this issue).
He has warned against the scourge of Equasy addiction, causing far more harm per exposure than ecstasy yet completely ignore by lawmakers. (
Equasy – a harmful addiction with implications for the current debate on drug harms Journal of Psychopharmacology, 23(1) (2009) 3–5 )
He has pushed for evidence-based harm scales for drugs, which of course do not fit with current legal scales. Especially since he doesn't treat tobacco and alcohol any differently from heroin and LSD. (David Nutt, Leslie A King, William Saulsbury, Colin Blakemore, Development of a rational scale to assess the harm of drugs of potential misuse, Lancet 2007; 369: 1047–53)
In one of the papers causing the flurry, Estimating Drug Harms: a risky business there is a very revealing table of how often deaths from different drugs are reported in the news:
|Drug Toxicological statistics||Newspaper reports||Toxicology to newspaper ratio|
(data is from Alasdair J.M Forsyth, Distorted? a quantitative exploration of drug fatality reports in the popular press, International Journal of Drug Policy, 12:5, p. 435-453 (2001))
While every death from ecstasy was reported in the press, none of the deaths from aspirin and just one in 265 from paracetamol were reported! Clearly it is big news that somebody dies from amphetamine or heroin, while diazepam (also known as Valium) deaths are uninteresting.
I wonder whether there is a harm reduction strategy for governments?