There was the Reinhart-Rogoff error, which claimed economic growth declines (0.1%) when a nation's debt passes 90% of GDP. For years this was used to set economic policy, especially austerity measures, in western governments. When someone (a grad student IIRC) actually bothered to check the Excel formulas, an error was found and the actual number should have been a 2.2% increase instead of a 0.1% decrease.
That the error occurred isn't amazing but it is mind blowing that so much government fiscal policy was set based in part on this one spreadsheet that no one bothered to double check. Blame extends both to those who used the model to justify the policies and to those who opposed the policy but never checked the assumptions it was based upon.
I feel like you're missing the most interesting point of this story - how many other policies are decided based upon nonexistent or erroneous data that nobody bothered to verify? You assume it's an anomaly, but what if this were the norm?
My post was in no way meant to be exhaustive but it's good to know that since you failed to list any of my other shortcomings in life that you believe me to be otherwise perfect. Afterall, you failed to list all of my failures so you must be assuming they don't exist.
My reasoning was that if you thought it were the norm, it wouldn't be worth mentioning. And you talk about it as a 'most interesting' example so it seems unlikely you didn't consider it an anomaly.
OP mentioned one error and the resulting policies because of it, I didn't see any others he mentioned. I don't think it's reasonable to know how many policies are determined based on sound data and processes, and how many are so politically motivated that nobody ever checks the numbers. You really only find out by happenstance when someone checks on a lark. We could make a concerted effort to verify everything, but where would you even begin for policies that have already existed for decades?
If I remember correctly, this wasn't really a "mathematics" mistake. The formula simply wasn't applied to the right number of cells (they didn't apply the cell-selection far enough).
Although that's beside the point. This was a tragedy of the academic economist profession, and should be covered day-one in any Econ research class.
I see the question as pertaining to pure mathematics, rather than engineering or applied science. For sheer mental anguish, I imagine it's hard to beat the flaw in Andrew Wiles's first edition of his Fermat proof. Discovering that must have sucked.
"First answer" is ill defined, since answers can move around depending on view options and later answers (and it's not chronologically first). Probably better to link directly: https://mathoverflow.net/a/9059 .
>All of the (in retrospect) misguided attempts to prove Euclid's Parallel Postulate, which eventually lead Gauss to develop hyperbolic geometry.
I had geometry last year and we mentioned that a lot of mathematicians tried to prove Euclid's Parallel Postulate. They tried adding a lot of properties to geometries. Saccheri is worth mentioning for contributing a lot in this development, especially for Saccheri quadrilateral!
That the error occurred isn't amazing but it is mind blowing that so much government fiscal policy was set based in part on this one spreadsheet that no one bothered to double check. Blame extends both to those who used the model to justify the policies and to those who opposed the policy but never checked the assumptions it was based upon.