House of cards
Some wild speculation and a bit of logic.
Can't be that easy to deliver such a large body of trained people. I expect that you'll have a large lsit of potential persons - there will have been some assumptions around how many of these expressions of interest remain worthwhile. Chances are that the more able applicants will have obtained work elsewhere and no longer be available.
What will be left is the dross - from which a larger percentage will have failed training or failed the CRB checks making the whole thing fail due to some over optimistic assumptions.
We are only finding out now as G4S will only invest their time and money in people who have passed CRB check and the training - no point in having a large pool of labour stood around doing nothing except costing G4S money until they are needed - and I guess these false assumptions probably made them feel safe to bet on submitting a bid 25% under the opposition.
I'm not trying to make excuses for G4S, nor am I connected with them - but in my experience some early miscalculations can lead to a set of assumptions (most likely in Excel) that create of house of cards. I don't know how you would test these assumptions either so you only find out its too late when its too late.