Liv Boeree to Break Up with Igor Because Math Says So!?!
7 years ago

27 Mar
Liv Boeree begins her TED Talk with an anecdote only those who have dated a degenerate will find truly recognisable, a cold hard assessment of how likely she and her boyfriend are to break up, complete with percentage chances. Her estimates werenât 0%, giving an unreasonable hope to most of her male friends, and the more optimistic sections of pokerâs online communities.
What she doesnât mention is whether or not she and Igor put any money put any money down on the over/under.
The somewhat frivolous tale makes a more serious point: exercises like this are what Boeree goes on to describe as a kind of emotional insurance â paying the premium of âan awkward conversationâ up front for some preparedness going forward. And her talk lays out some of the other ways in which our dealings with probability fail us, especially in unexamined aspects of language.
A Number Speaks A Thousand Words
The title of the talk is a little misleading because in among the range of points about language the central message of the talk is that numbers mean just one thing, with perfect conciseness, whereas phrases like âfair chanceâ and âalmost certainâ can mean a wide â and overlapping â range of percentages depending on context and individual preference.
The danger, according to Boeree is that using the wrong words can lead us into âgambling without realising that we are gamblingâ or to âtreating fringe cases similarlyâ. A fact she backs up with the provides the sphincter tightening statistics that the difference in a 99.99% and a 99.9% chance of a safety failure per nuclear reactor per year is the difference between one failure on average every 15 years and one every 18 months.
Experienced poker players are âalmost certainlyâ already using some of the ideas she lays out, but her proposals for habits that help generate the âclarity and granularityâ of statistical thinking in every aspect of oneâs life are worth the price of admission alone.
Watch the full talk here:






Comments
You need to be logged in to post a new comment