I was unclear, I was trying to point out that the sort of perks rachelbythebay used in her article were inconsequential with respect to the free cash flow, even giving Google the benefit of the doubt in terms of its costs by a huge margin.
When I was there and digging into this issue from the inside one of the issues was that the cafe's predict and prepare a given amount of food and that needs to be just a bit more than is actually consumed so that they don't run out. The remainder is waste, so their cost per meal has to include the cost over head of the waste. Also the preparation staff has a salary and benefits, and there is the cost of the kitchen gear (things you and I don't expressly call out in our own costs because we are the cooks and its our kitchen). But it is always safe to use the retail prices of restaurants in the area because they are demonstrably getting by on that level of return for their meals.
One of the counter arguments an executive made to me was that you couldn't predict the quarter, and I suggested you run it 'behind' so that Q4's free cash determined Q1's level of perks, that way you spent the money you had previously banked. He did not believe there was an accounting system that could deal with that sort of setup. I suggested that Google had one or two smart people who could probably figure it out. But the resolution was 'that would be too much trouble.'
And I completely agree that its their choice on how to do it. But here is the rub, really really smart people in the bay area can work anywhere. Literally walk out on friday, and before they have used up half of their accrued vacation they can be working somewhere else. So screwing around with this sort of stuff when you could afford not to, pisses them off and they leave (see rachelbythebay's comment about 'brain drain'). The secondary effect is that one of the reasons people give for wanting to work at Google is "all the really smart people there." except that the 'really really smart' start leaving, followed by the 'really smart' who start dealing with only the 'moderately smart' because their ranks are leaving.
Its an interesting chain reaction. And of course eventually all that excess money stops coming in because the people who were really responsible for it being there in the first place are now gone, and that leaves you no choice but to cut perks even further. The system often re-stabilizes at a much much lower productivity point.
I got to watch the whole sequence at Sun in the 10 years I was there, a friend of mine watched it at Oracle. When I joined Google I was optimistic that given that Eric Schmidt had been at Sun too that he also understood the dynamic that made technology companies great, vs just ok. I was mistaken.
The process is probably irreversible at this point. Google will coast on its reputation for a while, but folks inside will feel the pain of the 'missing ones', and things that used to just 'get done' will become big projects which take too long and too many people and are always behind schedule. They will keep lowering their hiring standards as the need to back fill positions of people leaving overwhelms their ability to get anything done. The ranks will swell with earnest folks who really want to do well but the ratio of smart people to earnest (but not smart) people will fall and so effort will increase without an increase in output. Revenue per employee will start to drop, and the days when 20,000 employees to get you $2B in free cash flow will be replaced by 50,000 employees and less than a billion.
I wish it were different, but sadly I don't see an out for them.