Perhaps they are maximizing earnings with an outlook of fifty to one hundred years. Allow their engineers a free hand so that product offerings are as diverse as possible and can later be integrated for advanced usages. Build interesting things and simply give away those not immediately profitable to gain and maintain constant mindshare in consumers. Rather than product ideas starting at the top and trickling down through layers of bureaucracy, anecdotes seem to tell of a flattened structure wherein thousands of independent projects churn in a chaotic maelstrom of ideas, with internal popularity and management approval selecting how projects will be made public, monetized upon or released through marketing machinery.

As to your example, free optical opens consumers to try new technology since their costs are minimal. This creates general consumer knowledge that the technology exists, which creates demand, which can be fulfilled by monetizing the persons already geared for running and maintaining it. And if the experiment fails, it was decent marketing and proved the market doesn't exist without having to gear up and release product specific sales, marketing and other auxiliary offices needed for a paid product launch.

One presumes that shareholders of google buy in with the knowledge and approval that the company works differently than many others. Whether this direction will be advantageous or disadvantageous is what the stockholders are wagering upon.