That doesn't track. As long as the project is generating at least a single dollar (TVM-adjusted) and you're not constrained on resources (SWEs, labs, etc.) that could be reassigned to higher yielding projects, that's all the justification you need.
The market tracks how profitable you are as company. Tech companies get higher multiples on their valuation because they can be more profitable than a manufacturing company.
This was one big reason why the trend is for conglomerates got broken apart, it lets their high profitability companies get valued higher.
Google focusing on products that are break even or slightly profitable hurts them for this very reason.
So if you are the CFO of google the decision you make is do we keep some of these low profitability companies around and have them drag our multiples down or do we cut them when its clear they won't become high margin profitable businesses.
Given that the CFO of google gets most of their compensation in stock its not surprising that they chose to have a higher multiples applied to them, and therefor higher stock prices, than lower ones.
...which partly-happened via licensing in the article, but they didn't realize the full value of the working operation.
Also, it may be slightly profitable when run under the umbrella, but not once you consider the cost of separate administration. That's especially true for things like Google Reader --- that project benefited enormously from living on Google Infra, behind Google Login, and you'd have to do a lot of refactoring and rebuilding to get it to run elsewhere, and then you'd have to get people to create new accounts, etc. And there wasn't much revenue there, certainly not enough to make it a viable standalone business.
But also more realistically, Google reports their margins to wall street and a ton of barely profitable ventures would drag that down.
The vibe I always get is that they won't hesitate to abandon stuff that doesn't get huge fast, but a big part is that the manager or teams don't want to get stuck with something that isn't a juggernaut or obviously on its way.
And heck, maybe that is just everywhere, but anecdotally at google, from an outside observer, it looks like the culture dictates being on the 'Big Thing' is how you succeed there.