« IBM and Philanthropy | Main | Business and Social Media »

December 12, 2011


I'm not sure what this reflects more negatively upon...the researchers, or the fact that they may be a proverbial canary in a coalmine.

As my first quant boss said: "90% of my ideas fail, and most people will tell you I'm pretty damn good at what I do."

And as Thomas Edison said: "I haven't failed--I've simply found 10,000 ways it doesn't work"...doesn't that mean that articles that say "I did A, B, and C, and my results sucked." have at least *SOME* meaning, if not giving someone else some idea, then at least showing what doesn't work?

After all, I once read Ed Thorp's "Beat the Dealer", and Thorp is basically regarded as the grandfather of all quant finance. And he got the idea because while studying for his PhD in some obscure nonsensical mathematical field at UCLA, he came upon a paper on Blackjack that failed to beat the house, but made the house advantage so close that it warranted further inspection. And after Thorp found a flaw in it (it constantly replaced the cards dealt), he went on a tunnel-visioned quest testing and re-testing ideas until he made money proverbially hand over fist in the casinos, and when the casinos started twisting and turning the game to take away a player's advantage (no card counting, no doubling down on some values), he took those same disciplines to the financial markets, and has been making a killing managing his own money ever since.

Think about that--if not for a paper published that was a complete failure, Thorp would never have done what he did.

I think the academic journals need to get off their high horse, realize that the purpose of articles is to advance human knowledge as opposed to have an ego contest, and work towards the benefit of mankind, not their own reputation that is only cared about by so few people.

This is a trove of good material. I hope you don't mind if I sit here for a while looking it over and thinking about how I can explain and use some of this for my science/technology writing class!

By the way, just sent you a question to Gmail.



In a session presented by Elsevier last year on how to write a world class paper, we were warned that a paper is typically read in detail by only one or two readers! I presume this does not include the reviewers. And a surprisingly large share (25-30%) of articles published even in leading journals with high impact factors are not cited within five years.

The slides from the very clear presentation (or a version of it) are available at http://www.docstoc.com/docs/52577859/HOW-TO-WRITE-A-WORLD-CLASS-PAPER

Thanks for the comments! Maurice - thanks a lot for the presentation. I hope that papers are typically read by more than 1-2 readers, but I suspect the real number isn't much higher. I'd never heard that up to 25-30% of published articles, including leading journals, are not cited within 5 years, but it makes sense that people would gravitate toward a small subset of papers. There is a snowball effect, I think, where people keep citing the same papers again and again. It certainly shows that dissemination of research doesn't stop at the paper being published.

The comments to this entry are closed.

On my shelves

Blog powered by Typepad

Creative Commons License

Aurelie on Twitter

Enter your email address:

Delivered by FeedBurner