Brooke's LawAdding programmers to a late project makes it later.

Risk control is executed by Ltrain's duck-and-cover rule:1. Bail-out before judgement day,2. Find an appropriate and gullable scapegoat,3. And if all else fails burn enough vacation time until it all blows over.

QuoteOriginally posted by: CuchulainnQuoteOriginally posted by: rmaxFor one-man projects I ask people to do as follows (because they are over-optimistic):a = optimistic durationb = realisticc = pessimisticthen expected = (a + 4b + c)/6This is for each activity in project; we do a risk analysis at the same time; developer must explain why he says it take 23 days work.For the total project, these estimations work out nice.My thesis advisor had what turned out to be the ultimate advice on calculating such things."Multiply by pi."If you're in over your head, pi^2 works.-Lugos

QuoteOriginally posted by: lugosQuoteOriginally posted by: CuchulainnQuoteOriginally posted by: rmaxFor one-man projects I ask people to do as follows (because they are over-optimistic):a = optimistic durationb = realisticc = pessimisticthen expected = (a + 4b + c)/6This is for each activity in project; we do a risk analysis at the same time; developer must explain why he says it take 23 days work.For the total project, these estimations work out nice.My thesis advisor had what turned out to be the ultimate advice on calculating such things."Multiply by pi."If you're in over your head, pi^2 works.-Lugos"multiply bu pi", that's great It allows for 300% project overrun and still have 14.15% pocket money left over.However, for fixed-price tendering it will not work.

A prime cause of project failure is that managers adopt a model based upon accountancy rather than the sort of basic risk management we all know and love.You need reserves to deal with fuckups, and to exploit opportunistic targets.When I reached the stage when I could state this as an explict assumption, I found that the sponsors of the project were keen.Meant I could say, if we see something really useful, we can grab it, and when something goes wrong, we don't have to sacrifice functionality and/or timescale.That typically meant contractors. I observe that quant development has few reserves. The nature of business untis is that they can't have 2 or 3 spare people lying around, yet the contractormarket is pathetically underdeveloped. That's partly my fault.I have failed to find the form of words thast will convince managers to take on temp quant developers.There are developer contractors, and some have quant skills, but they are typically used not as a reserve, but as a replacement for permanent staff due to politcal and/or headcount issues.