[sho_hn]

Even though Steam invites its own problems (it's ultimately a form of DRM, a closed platform, and a closed-source application), as a FOSS developer I can't help but feel excited about this prospect. I know many, many people in the 15-35 age bracket who are open to and curious about Linux, even tried it, but ultimately didn't stick with it because of the lack of high-end native games and because rebooting or setting up Wine is too much of a hassle. Valve has tremendous power to change this and legitimize Linux as a platform in their eyes.

Plus, there's already a lot of games in the Steam catalogue that have native Linux versions available:

- Dozens of independent titles, e.g. everything that was in those Humble Bundles.

- Everything using the DOSBox emulator to run even on Windows, e.g. id's Commander Keen, some Lucasarts Star Wars games, etc.

- Even a bunch of AAA titles: id Software's games (Doom, Quake) and games that have licensed their engine (e.g. Human Head's Prey), games that were ported by Linux Game Publishing (e.g. Egosoft's X series of spaceflight simulators), several games by Epic (e.g. Unreal Tournament) or using an Epic engine (e.g. Rune and Deus Ex, ported by Loki), Neverwinter Nights, Civilization: Call to Power, ...

Add Valve's own games and possibly some of the other games using their Source engine, and you could easily make 100-150 games available on Linux within a year of launch just from what's already there. But even more exciting is the notion of Steam's availability making more game makers consider adding Linux to their list of supported platforms going forward because the distribution problem is solved for them.

[toyg]

The distribution bit is fairly irrelevant -- yes, you have different packages and filesystem standards, but those are trivial and have mostly been solved. The real problem for Linux games is platform fragmentation: every distribution has a specific set of libraries running with a specific set of kernels, they change very quickly, everyone can compile them with whatever flag they feel like... Anyone distributing binary blobs is going to hurt in support terms, or make customers hurt trying to solve library riddles. Just look at how painful it is to try run an old Loki game on a new distro release.

[sho_hn]

No, the distribution bit is key and critical.

"OK, so we're going to do a Linux version. How are we going to get it into the hands of our customers? What, we need to set up a download infrastructure for that? No, that's way too much overhead for a Linux version. What, we're supposed to partner with a small niche digital distribution platform we've never heard about? No, not worth it, either. Wait, we can just upload the Linux version to Steam, where we already upload our Windows and Mac versions? Sure, why not."

As for distributing binary blobs: You're overstating the issue, IMHO. ABIs in Linux userland are fairly stable these days.

[toyg]

All this supposed stability doesn't seem to be reflected in a growing number of desktop products for Linux, not even ones that don't require a specific "download infrastructure" like recent games (which is really just a way of enforcing DRM, something that might not even be possible in the Linux world), so I'm really not convinced the main obstacle to commercial Linux development for the desktop is the distribution channel.

I think the bottom line is the opportunity price, and it's still way too high for Linux overall -- it's harder to support, harder to develop against, brings a several-orders-of-magnitude-smaller userbase who mostly resents having to pay for software... easy deployment won't change that figure very much, IMHO, but I'd be more than happy to be proven wrong.

[sho_hn]

> All this supposed stability doesn't seem to be reflected in a growing number of desktop products for Linux

You're shifting the goal posts. Nobody was claiming that ABI stability would lead to an increase in products, I was replying to you claiming ABI instability as a significant obstacle. As for the number of binary-distributed desktop products for Linux not having grown lately, I think you're completely wrong on that: See the Humble Bundle games.

And the userbase being so much smaller is, again, why Steam for Linux is such a big deal: It's an extension to a platform publishers are already familiar with, which makes it easier than ever for them to add the platform.

On top of that, Steam is more than just a distribution platform, it's also an SDK for things like multiplayer lobbies, meta-game systems like achievements and savegame sync. Dozens of titles are using that SDK now, and getting it ported to Linux maikes Steam for Linux also an important middleware port.

[peterb]

This is very forward thinking by Valve. It opens up the possibility of Linux-based game appliances which could run as VMs, dual-boot, consoles, etc.