Friday, June 19, 2015

science and policy #101

Three recent pieces of work in Cambridge came to light

1. scientists have been working on the basis for randomized trials, and realized that, of course, we must have some non-randomized trials, to check if the very basis for randomization as part of scientific empirical method is sound.
In a bold inter-disciplinary move, the scientists collaborated with the department of history and analyzed a number of UK and other policies for economics and military action, to see if one could find random (e.g. the 100 years) and non-random (e.g. the 1st world) wars, as well as economics (e.g. monetarism, and austerity). The results will be published very soon, but are currently under embargo, in case they disturb a current experiment with Greece.

2. Engineers in Cambridge have long wanted to build a railway to replace the ageing bus and taxi system. Working from earlier chinese experiments with mono-rails, and the guided by the guided bus success, the proposal is not to take the modern electric line from Royston to King's Lynn, where customers are already used to the trains splitting at Cambridge, with one half going forward, for example, to Ely, and the other half, soon, to the Science Park. From next year, they hope to split the train laterally, with the left half going around the pieces (Christs, Parkers) and Commons (Midsummer etc), and the right half going in a long overhead loop, to Ely, allowing the Eels much easier migration along their breeding paths in the fens. If the duo-mono-rail is a success, the engineers propose to extend the routes to Paris and Brussels, where onward mono-mono routes could serve ski-resorts and some of the Belgian mountain regions where the finer beers are produced.

3. For some time now, a very ambitious project in CRASSH has been working on Consipiracy Theory. This work has involved linguists, computer scientists, taxi drivers and publicans, and has recently yielded a breakthrough. A new tool has been built that can detect consipiracy theories with a false positive rate of 2% and a false negative rate of 3%. The method is based on a mix of Bayes and various NLP clustering algorithms. Currently the tool is part of a possible startup and venture capitalists are clamouring to fund the work. The business case is unclear as yet, and there have been some suggestions that at least one major journalism organisation may have prior art, although scientists suggest that their conspiracy generator is based on different technology (followers of Chomsky will understand that recognition and generation are quite different linguistic machines). At least one government agency claims that they had build a system exactly like this in 1961, and that it correctly identified Cuba and Suez, but they could not reveal the technology for fear of showing potential national enemies how much more advanced the UK was than them. Security analysts have asked them to "put up or shut up" as this is not the first time that they have claimed to have approaches to their work that would save time and money, but have not deployed because they would have, err, saved time and money and lives and red faces.

Meanwhile, CRASSH were not available for comment.

No comments:

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