How Many LHC Failures Is Too Many?

Re­cently the Large Hadron Col­lider was dam­aged by a me­chan­i­cal failure. This re­quires the col­lider to be warmed up, re­paired, and then cooled down again, so we’re look­ing at a two-month de­lay.

Inevitably, many com­menters said, “An­thropic prin­ci­ple! If the LHC had worked, it would have pro­duced a black hole or strangelet or vac­uum failure, and we wouldn’t be here!”

This re­mark may be some­what pre­ma­ture, since I don’t think we’re yet at the point in time when the LHC would have started pro­duc­ing col­li­sions if not for this malfunc­tion. How­ever, a few weeks(?) from now, the “An­thropic!” hy­poth­e­sis will start to make sense, as­sum­ing it can make sense at all. (Does this mean we can fore­see ex­e­cut­ing a fu­ture prob­a­bil­ity up­date, but can’t go ahead and up­date now?)

As you know, I don’t spend much time wor­ry­ing about the Large Hadron Col­lider when I’ve got much larger ex­is­ten­tial-risk-fish to fry. How­ever, there’s an ex­er­cise in prob­a­bil­ity the­ory (which I first picked up from E.T. Jaynes) along the lines of, “How many times does a coin have to come up heads be­fore you be­lieve the coin is fixed?” This tells you how low your prior prob­a­bil­ity is for the hy­poth­e­sis. If a coin comes up heads only twice, that’s definitely not a good rea­son to be­lieve it’s fixed, un­less you already sus­pected from the be­gin­ning. But if it comes up heads 100 times, it’s tak­ing you too long to no­tice.

So—tak­ing into ac­count the pre­vi­ous can­cel­la­tion of the Su­per­con­duct­ing Su­per­col­lider (SSC) - how many times does the LHC have to fail be­fore you’ll start con­sid­er­ing an an­thropic ex­pla­na­tion? 10? 20? 50?

After ob­serv­ing em­piri­cally that the LHC had failed 100 times in a row, would you en­dorse a policy of keep­ing the LHC pow­ered up, but try­ing to fire it again only in the event of, say, nu­clear ter­ror­ism or a global eco­nomic crash?