Astrology & the 32 bit integer

Firstly, hats off to Dstyle who was, I think, the first to spot early on the uncanny resemblance between the amount put up at 29.00 and the maximum value of a 32 bit integer.

This of course is only the beginning of the story.  There are still many unanswered questions that to my mind go to the very heart of the integrity of the Exchange.

What do we know so far?

Very little, other than the Betfair statement which included the following:

"In this case the customer had less than £1,000 in their account so none of these bets should have been accepted. However, due to a technical glitch within the core exchange database one of the bets evaded the prevention system and was shown on the site. This was an issue that was triggered because of a unique sequence of events that had never happened before."

We also have acting CEO Morana reported in the Guardian as follows:

"When we investigated, it was as if there had been 10 different stars all in alignment," Morana said. "I would have said before the Leopardstown race that this was something that could never happen, the odds against it were just infinitesimally large.”

Discussions and questions

Leaving aside the implied minor nature of the technical problem (Webster: Definition of GLITCH. 1. a : a usually minor malfunction …; also :bug 2 b : a minor problem that causes a temporary setback : ...), what could have been wrong “within the database” that allowed the bet through, even with integer overflow?

Equally open to question is the Lemony Snickett defence (A Series of Unfortunate Events): variously described as “a unique sequence of events that had never happened before”, or as an astrological rarity - “it was as if there had been 10 different stars all in alignment”.

Ignoring the fact that it is planets not stars that change their alignment (well, certainly not meaningfully in Betfair’s lifetime), I feel this merits further exploration.  As an old buffer I’ve lost track of the number of times over the years someone from the IT department has come up to my desk and tried to explain the latest system failure by claiming it was a one in a million happenstance.  Morana said “the odds against it were just infinitesimally large.”  Someone, I think Tony Calvin, said that Betfair had processed over half a billion bets since it started.  So can we take it that “infinitesimally large” is many times greater than 500,000,000,000/1?  If not then this was an accident waiting to happen.

This begs further questions in my mind.  What other aspects of Betfair’s critical functions (bet placement, cross-matching, accounting, etc.) have been built with the same philosophy – if it’s highly unlikely to happen don’t worry too much about it?  Could they not have built this thing using serial logic, rather than relying on the odds of something happening or not happening?

Given the evident philosophy and the very high transaction volumes, can we be confident that other unusual astrological “infinitesimally” remote circumstances have not already happened (perhaps undetected, or unnoticed), and will not happen in future?

I am with the many already out there who are rightly in my view demanding far greater exposure of the technical details of what happened.  It is essential to restoring trust, which is a key pillar upon which this exchange is built.