I said: > I dearly remember a situation where > I had to defend a Monte Carlo simulation > based on three draws. Anyone who knows > some about Monte Carlo simulations knows > that you generally need many thousands of > draws to get any meaningful result.
After I wrote this, I remembered another experience. The very first paper I published was based on some numerical technique and after it got published, I discovered a mistake in my program. It was an unintentional mistake so I was not that much worried. We all make mistakes. But I know from my experience with about a dozen numerical papers that none of my referees reviewed the codes I used to obtain my results. Now, in many academic areas ranging from numerical fluid dynamics to econometrics, people produce numerical results and get them published. Of course, we all expect that some of them will have mistakes in them, hence there is no problem so far. On the other hand, over the past few decades, and, particularly over the past decade, the American academe started to pay huge amounts of money to its stars. Stars are everywhere from physics to engineering to economics to what have you. These stars are under heavy pressure to produce "high quality" work all the time. On top of that, there is this heavy publication pressure on the young faculty to get tenure. How do we know that none of them submit some of their papers with mistakes that are known to them? Stakes are high, aren't they? Sabri