just because something starts as a free photosharing app doesn't mean it stays that way. One of pg's main points is that entrepreneurs see a way in that is often undervalued by others for a variety of reasons. When MSFT came along people underestimated the value of the OS, for Apple it was the PC, in Intel's second coming (first being memory) it was the microprocessor which even Intel itself underestimated for a while. Before Apple's rebirth phones and music players were seen as a small, relatively commoditized business. I dare say that a large part of FB is a free photo sharing app...
No, PG states a preference for revenue growth... with user growth being an acceptable stand-in, contingent on the usual assumption that users will convert to revenues.
And the VCs in his model are not the end goal, but only a middle stage, to accelerate growth and thus dominate a market, creating an enterprise that will be valuable to others on some combination of traditional factors (such as discounted expected profits or synergistic/strategic value when combined with an existing business).
I think you're trying so hard to see what you already believe you're missing parts of the PG argument.
pg explicitly wrote that active users are a proxy for revenue (in an early stage startup). In the early stages of your business, measuring revenue is hard because there is not enough data points yet. But you need to measure something that approximates your potential revenue to judge impact of your actions.
First, let's accept that freemium is a good honest business model that by its nature is a good fit for many Internet services. If you build freemium model similar to DropBox, conversion to paid users is typically around 1-10% from the active users. Thus, it doesn't necessarily make sense to introduce a paid plan and implement integration to payment systems (a task which was PITA before Stripe) until you have a way over 10k active users. Furthermore, something like 30k active users might mean that you need over 300k signups, which will take a while if you don't have explosive takeoff. Thus it makes sense to approximate potential revenue by measuring active users, when your actual revenue stream is still not fully in action.
As I said in another comment, the same approximation philosophy works in enterprise sales too. You measure the performance of your sales pipeline, which works as a proxy for revenue even before you close your first $100k deal.
We could have said the same thing about search engines. Banner ads are never going to pay the costs for all that traffic and all the needed engineers...
