Interesting blog, showing the banking-side view. The call for 'executable-regulation' reminded me that lawyers also pride themselves on their fine analytical minds, piling up sharp clauses to cover this and that.
I suspect that, if they could, and would, run unit tests on their products, there would be much wailing and gnashing of teeth.
On the other hand, we already have that with the current system, and that's perhaps the point of it. The making of sausages, and all that.
" The idea that while ABS documents are chock full of legalese, a computer program can provide a very concise way of understanding how a financial instrument is supposed to operate."
That is extremely interesting. A good friend of mine is a lawyer and used to work on large financial contracts (with sometimes billions of dollars in a single contract). He would tell me crazy stories about how two legal teams would spend hours arguing over things like the placement of a semicolon. So in reality legal contracts really are code...just a code for lawyers and not computers. I would not be surprised in the least if computer code was more efficient, but it may take just as long to write (if not longer). The upside is that while lawyer code needs to be tested in court, computer code gets tested by compilers and private QA teams, significantly shortening the process and relieving a significant portion of bureaucracy.
I forgot to mention it in the blog post, but there are well defined measures of code coverage for software development. I'm not sure if the same concepts have made there way into legal analysis. I would like to see that any financial portfolio 'stress tests' require complete code coverage.
However, I seriously doubt whether the proposal will ever be enacted. There is just too much going against it. I do like that agent simulation is becoming wider known in the broader financial community as it finally gets us closer to acknowledging that financial instruments _are_ computation. And along with that association comes the tacit recognition that not everything is easily computable.
Velupillai's book is awesome. An exercise of his that I love is looking at how much of general equilibrium theory one can prove using only constructive math.
A nice blog on a similar vein seems to have been started recentely on http://aiecon.tumblr.com/ by a grad student of Sandholm.
Thanks for that. Though I'll note that it's a bit cruel mousing over the "full text" buttons, seeing a "You have access to the full text for this item" balooon/tooltip, and, upon clicking on the Preface, getting a 'Log In' page ...
5 comments
[ 6.8 ms ] story [ 32.7 ms ] threadI suspect that, if they could, and would, run unit tests on their products, there would be much wailing and gnashing of teeth.
On the other hand, we already have that with the current system, and that's perhaps the point of it. The making of sausages, and all that.
That is extremely interesting. A good friend of mine is a lawyer and used to work on large financial contracts (with sometimes billions of dollars in a single contract). He would tell me crazy stories about how two legal teams would spend hours arguing over things like the placement of a semicolon. So in reality legal contracts really are code...just a code for lawyers and not computers. I would not be surprised in the least if computer code was more efficient, but it may take just as long to write (if not longer). The upside is that while lawyer code needs to be tested in court, computer code gets tested by compilers and private QA teams, significantly shortening the process and relieving a significant portion of bureaucracy.
However, I seriously doubt whether the proposal will ever be enacted. There is just too much going against it. I do like that agent simulation is becoming wider known in the broader financial community as it finally gets us closer to acknowledging that financial instruments _are_ computation. And along with that association comes the tacit recognition that not everything is easily computable.
For a good introduction to the intersection of computability and economics, check out http://www.oxfordscholarship.com/oso/public/content/economic...
A nice blog on a similar vein seems to have been started recentely on http://aiecon.tumblr.com/ by a grad student of Sandholm.
Anyways, Amazon seems to have it, and Google Books lets one have a peak, at http://books.google.com/books?id=LDmWrvPrOvUC&dq
More to the point, Book Depository, while listing that 2000 book as out of stock, does have an interesting looking 2009 Computable Foundations for Economics, by K. Vela Velupillai at http://www.bookdepository.co.uk/book/9780415355674/Computabl...