The Steam Review

Gamesindustry.biz have posted an interview with Mark Morris, director of Introversion Software, developers of Darwinia and Defcon, which spends a significant portion of its time covering the indie developers’ relationship with Steam, and their thoughts on digital distribution in general.

Summary of points

• Introversion contacted Valve on a whim, and found that working with them is “much nicer” than with publishers thanks to Steam’s direct nature. More
  

  We were prompted to [distribute on Steam] when Alison [Beasley, from Introversion’s PR firm Lincoln Beasley] said, just after the Edinburgh festival, “You should talk to Valve about going on to Steam” – so we did. TJ got in contact with Scott [Lynch] over there and showed them the product, asked if they were interested…they were, and they took it, and it’s had pretty good initial sales, which has raised our hopes. It’s gone down a bit since then, but it’s prompted us to do digital content on all sorts of things.

  It’s much nicer to work with Valve than it is to work with some of the big publishers, where you’re so far away from your customers – with Valve, it’s you, Valve and your customers, which is great.

• Although they have no fixed contract, Introversion are confident that their future games, including Defcon, will be made available on Steam in the same manner as Darwinia. More
  

  We don’t really want to sign deals and say, “right, the next four titles are going to go out over Steam,” because we don’t know what’s going to happen. At the moment the relationship with Steam is really good, and I’d love to have just put title after title after title out on Steam. We’re talking with them about Defcon, and I’m confident that we’re going to get Defcon going over Steam as well.

• Introversion are arranging an Xbox Live Arcade release for Darwinia. More
  

  We’ve got Steam on the PC retail, hopefully Live Arcade – we’re talking to Microsoft, so hopefully we can work that out.

• Valve don’t seem too sure what their strategy is for the growth of the Steam library. Some staff will tell you that there is a plan that “will become clearer over the next year or so”; others that anything they consider worthy will be put up. More
  

  I was having a chat with their new products guy, the guy who’s responsible for what goes onto Steam, and there are differences in opinion within the company. Some people will tell you, “we’re going to put every game on Steam – all these things are going to be on there”; other people will say “no, we’ve got a very clear idea and a strategy of the sorts of games we want to have on Steam, and that strategy will become clearer over the next year or so.”

  I don’t know whether it’s the case that they do have this big master-plan, and Darwinia fits into it, or whether they’re just actively looking for content, and there’s not a huge amount of content out there at the moment because people are already signed up with exclusive deals with box copy publishers and all the rest of it. So, I don’t know, but it’s an interesting question alright.

Valve’s library strategy

Introversion’s third game, Defcon.

Morris’ fourth quote is particularly intriguing. What exactly is Valve’s strategy? Edge magazine reported that Valve were “all but trawling the IGF booths with a shopping trolley” in their GDC 2006 write-up (thanks hahnchen), yet the volume of titles present today does not suggest a proactive search. All we know is that casual games are not accepted.

There are three key-frame approaches that can be taken for online library growth. The first is accepting absolutely anything that comes your way; Triton appears to be closest to this method. It takes advantage of the theoretical “infinite shelf-space” of digital distribution and increases your target demographic, but very much at the expense of your image. It isn’t the subtlest of options.

Read the rest of this entry »

I recently upgraded to the Bad Behaviour 2 anti-spam plug-in. If you are having troubles commenting or searching, please tell me!

Triton was quietly released at E3 this year, though as a “soft launch” nothing really changed except the removal of its beta tags and the sale of a handful of disinteresting budget titles. The system is now starting to move into top gear however, with this week’s launch of the Prey pre-load (thanks Sarkie) starting the push.

Triton’s interface has undergone improvement, but is still clunky.

Yes, pre-load: DiStream have sensibly followed Valve’s lead and are offering an encrypted download of the game’s content ahead of its July 11 release. Unlike Valve’s solution however, Triton pre-loaders must also be pre-purchasers; the download will not commence without the billing of the user’s credit card. Perhaps this is a bandwidth-saving move, but for my part I believe that it is to ensure that the inevitable system bottlenecks and bugs do not overwhelm Triton during its first public trial. Certainly, it is an approach that improves on the various baptisms of fire weathered by Valve and their systems.

But for all the benefits requesting card details up-front brings, the issue still remains of enticing users to hand them over. When you pre-order a game through Steam the chances are you already have it safe and secure on your system, and that leads to several subtle psychological effects. Are you really going to let it sit there and rot? Why not slap in your 16-digit string and make your time worthwhile? When you have nothing but a handful of screenshots and a PR blurb in reach, making the leap is much harder.

Add to that the fact that most will never have heard of Triton before seeing its logo in the Prey demo’s nag screen, and an interface that, while having being greatly improved from its original incarnation, is still bulky and inefficient, and you have a recipe for low uptake. Despite my knowledge of DiStream and Triton, I too am finding myself somewhat apprehensive of making the purchase, which comes to £27.35 or $49.95.

The price, like Half-Life 2‘s, is similar to that of the game’s retail release, though probably for different reasons. Valve resolved the issue of enticing buyers without undercutting retailers with it’s Bronze, Silver and Gold packages, but with only one game Human Head and DiStream do not have that luxury, piling on yet another obstacle.

Like Valve, DiStream are unlikely to release sales figures. In all but the most extreme of outcomes we will only be able to speculate on the service’s success: right now, it looks like things are tipping towards under-achievement.

To the public, the Steam API currently consists of metadata for mod teams. A “future” Source SDK update will extend it with a Friends SDK for chat plug-ins, but any frameworks for plug-ins to Steam itself remain shrouded. Which made the announcement that the joint CAL-Verizon FiOS Grand Tournament was to be managed through a third party client distributed through Steam, Playlinc, of great interest. What would Playlinc’s “new tricks” be? What would Verizon do to tap Steam’s potential?

The answer to that last question, we now know, is absolutely nothing.

This isn’t as unreasonable as it might seem. Although the chasm that exists between Steam and Playlinc extends to ludicrous extremes, to the point at which tournament members are required to enter manually the results of each match, even when each one takes place on tournament servers, and at which to use the client in the first place one must hold an ICQ or AIM account, there are several factors which could be preventing greater integration, and several more that probably are.

Foremost, Playlinc is an implementation of Orion, which was likely never designed with integration with anything but games in mind. Equally, Steam inflexibilities cannot be ruled out – perhaps a plug-in framework was to be written for the tournament but was not finalised in time. We certainly know what Friends only left its extended beta at the end of May. These things do not excuse the lack of server stat logging, but, in addition to Playlinc’s future as a standalone client, would explain why so much potential has been sidestepped. While it appears ridiculous to demand the use of an external chat service when an internal one is already available and, crucially, extends in-game, it would seem that Verizon have a solid set of excuses.

Elsewhere in the tournament, integration has been more successful. Online events are usually severely limited affairs requiring dedicated clans and strict timetables, and it is only digital distribution and Steam that has allowed FiOS Grand to open itself up to anyone who felt like entering, even if they did not own the game. Playlinc itself allows the dynamic launching of secure servers on Verizon’s hardware, its key feature in fact, rounding off the coupling.

Perhaps the most important piece of wisdom we can take away from FiOS is that it has given concept of plug-ins a boost, no matter how minor it may or may not be. Next time the situation will hopefully be more pliable, and in all likelihood the relevant parts of the Steam API better established. Then we’ll see what can really be achieved.

Virgin Games is set to see the launch of a proprietary digital distribution service next year, delivering a range of titles from FPS and RTS to casual. No mention is made of the service becoming a full platform akin to Steam or offering anything other than downloads, but those downloads will have an interesting twist: they will be completely streamed. In the words of VG’s chairman Simon Burridge, “consumers [will] never have the full game downloaded”.

VG’s ephemeral model follows the same concept as the original design for Steam’s Distributed File System. Rather than keeping the entire game on the hard drive, only the content in use, and either in a set buffer zone or that doesn’t make the cache exceed its maximum size, is available locally. The idea was dropped from Steam when file sizes, rates of game production and connection speeds made it inefficient: it must therefore be asked how Virgin intend to distribute today’s hefty games through such a system without infuriating its users, and what sort of limits they will have to enforce to avoid bottlenecks.

That said, it is notable that the age range of the games the system aims for is not mentioned, and the very fact that the streaming system is ephemeral suggests that its developers are expecting a high turnover rate among games.

Technical hurdles aside, Virgin Games intend to use their system to deliver the first major DD service using a rental model. They believe that by ensuring that users never have an entire game on their systems the security issues with rent systems on PC that have held the process back can be circumvented, though again, it is unclear exactly how the system will help the situation. There is a sweet spot to be found between preventable casual piracy and the gormless and untiring number robots of cracking rings, and from what we know today trying to get files that are on a user’s computer off at a later time does not seem to hit it.

Nevertheless, if Virgin can pull the system off, more power to them. If successful it will act as a testing ground for subscription-based purchasing of AAA titles, and could well propel similar systems on Steam and other services forward, or forewarn them of the consumer backlash against the idea – or even both. A healthy dose of scepticism is recommended from this quarter however, as the system, due in the second quarter of 2007, is still in its infancy.

Steam’s virtual drive has long been known to erode the loading performance of games using it. A new filesystem beta started last night containing Valve’s solution: accepting the inevitable and leaving GCFs for the full system.

Games using what was formerly the Steam virtual drive now have their files loose on the hard drive in /common, a subfolder of /SteamApps. Small NCF indexes take the place of GCFs to ensure that the standard filesystem’s features (alternate content mapping, “the various Steam APIs” and streaming) can still operate.

John Cook explains:

Any virtual file driver causes an extra cross-process synchronization to occur, which although small by itself can add up to a lot of time when you have tens of thousands of read requests. The quickest thing that can be done is simply not doing that work. Going forward we have two strategies for SteamFS:

• Direct SteamAPI file access (GCF), which lets the app take advantage of the cache file format’s strengths (lots of small, easily updatable files in a compact format).
• Loose files on disk (NCF), which lets old or licensed games run on Steam without integration work or any unnecessary indirections.

Once we realized that we could still support our Steam features while having loose files on disk, it became an easy choice to do it for old games.

The changes bring the GamersFirst programme, overshadowing so many of Steam’s public developments recently, one step closer to availability.

The virtual drive’s demise also opens up WINE/Cedega emulation for games that do not implement the standard system.

Also, please note that the standard GCF implementation does not suffer from the latency of the virtual drive. The virtual drive is an extra layer above the Windows filesystem; the full GCF implementation replaces it.