Valve-watching is a sport that requires patience and a good eye. Sometimes years will pass without any discernible sign of movement. Until by some divine insight, the monolith steps forward and plants its foot in precisely the right place.
Except it’s not divine insight - it’s the result of masses of research. Let’s use an example. Dota 2’s beta was a daunting prospect for MOBA-babies Valve, and so they used TF2 updates prior to Dota's announcement in 2010 to test their ideas.
“We try to use our previous works’ tools to help us reduce risk from our future things,” explained TF2 creator Robin Walker on the Nerdist podcast. “There were things we thought we might want to do in Dota 2 long before we even announced the product, and we were able to look at existing products like Team Fortress 2 and do things in those products that were largely about gathering data so that we would know whether we should do them in Dota - and if we did do them in Dota, what were the characteristics [it should have] and the pitfalls we should avoid.”
At this point, boss Gabe Newell stepped in to clarify. “So a lot of the things that got shipped as updates in TF2 were to answer questions that we were having about what Dota 2 should be,” he said. “Rather than wait three years to find out which of our ideas were good ideas and which of them were were bad ideas, we can test them a lot sooner and prune out the bad ones a lot faster.”
TF2 is more or less the most successful example of game-as-ongoing-service I can think of this morning - more power to Valve for making the most of their metrics and improving both games simultaneously. Here’s a fun thought though: what does TF2 do that Dota does better?