"So the right way to do that is to make Steam essentially a network API that anyone can call. Now, this is separate from issues about viruses and malware. But essentially, it's like, anyone can use Steam as a sort of a distribution and replication mechanism."
The common refrain I hear is that Steam hosting user-created stores will negate the curated experience that has made it so successful. This misses the scope of what Gabe Newell is proposing. I don't think Steam will host user-created stores at all.
Steam as we know it today is made up of two halves: Valve's Steam Store, and the Steam Library. A customer purchases a game from the Store and (automatically) registers it with their Library.
I believe what Newell is proposing is that Valve will open up the first half, allowing anyone to develop their own standalone store using the Steam API. These stores will be external to the Steam Store and website. In fact, the Steam app itself will just be another consumer of the API. When a customer purchases a game from one of these stores, it will register with their Steam Library. This part will continue to be managed by Valve and integrated into the Steam app exclusively.
With this approach, Valve's Steam Store will continue to be a curated experience. But the Steam Store expands from being a single storefront into being a service, and extends its reach as a result. It's a natural progression of Valve allowing developers to bundle Steam Keys with non-Steam purchases.
Developers (and other curators) can use Steam's Store API instead of building a website or standalone app from scratch. They can take advantage of the payments, authentication, Steamworks, hosting and auto-updating infrastructure. This is a huge burden off developers, and a new opportunity for curators.
Customers clearly love using Steam, but I think it's less about the store itself, and more about the security of Valve's payments system, and the convenience of the Steam Library. The Steam API may not help with discovery, but it will make converting customer interest into purchases dramatically easier.
So this isn't necessarily the "Steam to riches" magic bullet. Mainly because these user-created stores will not be displayed in Valve's Steam Store. The challenge for indie developers will be, as it always has been, promoting and marketing their games. Valve will obviously invite the best developers into their store, but the bar isn't getting any lower than it is today.
A Steam API for standalone stores clearly benefits all three parties here: it expands Steam's (and subsequently, Valve's) reach without compromising their curated store business model. It makes developers lives much easier, and customers get a familiar, safe and convenient way to buy (even more) videogames.