GD Column 20: The Coming Storm

The following was published in the Feb 2012 issue of Game Developer magazine…

Ever since OnLive’s dramatic public unveiling at GDC 2009, the games industry has been watching and wondering about cloud gaming, at times skeptically, at times hopefully. The technology holds the potential to revolutionize the business, perhaps forever destroying the triangle that connects consumers with hardware manufacturers and software retailers.

Some of the immediate benefits are obvious. Instant, time-limited demos would allow every developer to showcase their games on-demand with no extra work. Frictionless per-day, or even per-hour, rentals would bypass Blockbuster and other rental chains, ensuring that more money goes directly to the people who actually make the games. Similarly, virtual ownership handicaps GameStop’s ability to resell a single disc multiple times, again making sure money flows directly from the consumer to the developer. Further, if a publisher commits fully to the cloud – with no offline version available – piracy would be virtually impossible

As for the consumer, cloud gaming enables cutting-edge graphics on any connected device, with no installing or patching ever. Although the system requires a constant Internet connection, more and more games – even single-player ones – are demanding a connection anyway. Indeed, cloud gaming can handle Internet stutters better than local gaming; in Diablo III, a dropped connection sends the player back to a checkpoint screen while OnLive returns the player to the exact frame she last encountered.

Most importantly for consumers, cloud gaming should change the economics of pricing. By removing the traditional retail middlemen, not to mention secondary drags on the system like rental and used-game sales, a developer could easily make as much money selling a game for $30 via the cloud as they could selling it for $60 via a traditional retailer. The industry could finally approach a mainstream price point, with games priced comparably to movies, books, and music – instead of the $60 price point (for a $300 console) which is absurdly out of reach of the average consumer.

Indeed, the economics could change for developers too. If entirely new business models emerge, with consumers paying for a game daily, weekly, or monthly – or perhaps with a single subscription to all available games (a la Netflix)  – the design incentives change. Cloud gaming could reward developers for depth of gameplay over ornate, scripted sequences; infinitely replayable dynamic games like Left4Dead or StarCraft might suddenly be more profitable than hand-crafted semi-movies like Call of Duty or Uncharted.

Rethinking Consoles

Cloud gaming also has important implications for the next generation of consoles. The ability to run games from the cloud gives the console makers a profitable alternative to both the rental market and the retail middlemen, all within their own closed systems. As a bonus, they could even sell inexpensive “cloud-only” versions of their next-gen console, without optical drives or hard disks.

Going down this path, however, raises the thorny question of whether consoles are even necessary at all. OnLive is already selling a “MicroConsole” that provides a current-gen console experience via the cloud; there is no reason similar technology can’t be included by default in new TVs, or even in any cable box or satellite receiver. What defines a console, after all? The three necessary elements are the controller, the screen, and the couch. Soon, anyone with those three things and an Internet connection to the cloud will be seconds away from any game.

Indeed, because cloud servers already dwarf current-gen consoles in horsepower, they can bring a next-gen experience to consumers today, by default. The cloud promises a “perpetual” virtual console that get updated regularly as new, faster servers come online. Publishers should be receptive as a perpetual console promises to end the boom-bust cycle they experience with each new generation.

Indeed, this year in the current generation’s lifecycle should be when publishers are making record profits. Instead, many are fighting just to stay in business; the next wave of forced upgrades could wipe them out. Anything that could prevent the gap years when consumers are forced to migrate between consoles would be a welcome change.

Thus, for the console makers, cloud gaming’s promise is mercurial – it could break them free from the parasitic drag of traditional retail but it could also destroy them by making the hardware itself irrelevant. The best defense against the latter is an active and direct relationship with the consumer, not tied to any one machine.

On this front, Microsoft, with its comprehensive Live service, is far ahead of Sony and Nintendo – many gamers would be hesitant to leave behind their gamerscores, achievements, friends lists, and downloadable games for another ecosytem. However, a radical step could cement this bond.

Because a thin video-based client can run on almost anything, any one of the console manufacturers could start the next generation tomorrow by simply buying OnLive or Gaikai and embedding it in the next system update. Next, they could sell cloud-only versions of the current-gen consoles for almost nothing ($100? $50?), which could revolutionize the market and inoculate the company from the coming shift. Some are predicting the next generation of consoles will be the last one, but it may not even be necessary at all.

Rethinking Games

However, cloud gaming’s potential is much, much greater than changing the economics of the industry; in fact, it could revolutionized the very way games are made. For starters, the cloud could solve the number one problem that plagues most teams – a lack of feedback from real players during the early stages of development when radical change is still possible. Most game project grow slowly from fast and nimble speed boats to hulking battleships which can only change course at great effort and cost.

Using cloud technologies, a team could expose its game to fans as soon as it is playable, with almost no technological hurdles or security concerns. All players need is a browser and, if necessary, a password. Releasing games early for feedback and buzz is nothing new for indie developers (indeed, doing so is their major competitive advantage); nonetheless, for major publishers, the idea is fraught with potential risk, of leaked games and bad press.

However, as the game’s code and assets would exist only on the cloud’s servers, nothing could be leaked. As for pre-release buzz, the greatest danger, of course, is of simply releasing a bad game, and the surest way to do so is to isolate a development team from the oxygen of real players. Further, the cloud’s inherent flexibility creates myriad ways to target players for testing: 24-hour passes, geo-locked sessions, early press versions, pack-in codes, and so on.

Further, the cloud promises more from these early test than some simple metrics or private forum comments. Because the output of a cloud server is a video feed, developers would have access to a recording of every minute ever played of their game. Wonder how players are handling a certain tricky boss? The designer can simply watch saved videos of many different players tackling the encounter.

Still, the greatest change cloud gaming could bring is the end of client/server architecture. Many online games have thin clients, with the “real” calculations being done on the server, largely to prevent rampant cheating. (As Raph Koster famously put it, “The client is in the hands of the enemy.”) With cloud gaming, the client is so thin that it might be inappropriate to even call it a client; it’s simply a video player that takes input.

The upside of this system is that developers would no longer need to waste resources developing a traditional game client, plugging its security holes, worrying about peer-to-peer connectivity, and optimizing what minimal, yet necessary, set of data needs to be sent to the client. In other words, making a game multi-player would now be essentially trivial.

Writing multi-player games is a formidable challenge – keeping game state in-sync between servers and clients in a safe, fair, and accurate manner is no small feat. With cloud gaming, these issues evaporate because there are no clients anymore. Developers simply write one version of the game, run it on a single machine, and update it based on user actions – which is how single-player games are made.

Taking advantage of this feature would require some courage as the developer would need to go all-in on cloud technology. Developing an online game with no client means that the game could only be played via the cloud. There are many benefits to being cloud-only – no piracy, for one – but the greatest benefit might be not even needing a network programmer.

Perhaps the group which has the most to gain from this new model would be small, independent developers, for whom the idea of building an indie MMO seems laughable given their tiny resources. One can’t help wonder what Mojang could do with a cloud-based version of Minecraft, seamlessly updated, playable from any device or browser, that connects every world end-to-end. So far, the big question for cloud gaming is when will it be feasible, but ultimately, the more important question is what will it enable.

8 thoughts on “GD Column 20: The Coming Storm

  1. Wow!Great article.I had no idea there was such an upside to cloud gaming.Especially given the fact that it would make it easier for small developers to stay afloat.

  2. “cloud servers already dwarf current-gen consoles in horsepower”
    Certainly in RAM amd bandwidth, maybe in CPU, definitely not in GPU. And if you consider that multiple users are supposed to share a single streaming server, it’s not that obvious anymore. Do you have the Gaikai or OnLive server specs? Do you know how many users are hosted on each server? (I don’t)

    “Wonder how players are handling a certain tricky boss? The designer can simply watch saved videos of many different players tackling the encounter.”
    How many hours per day would designers spend on that? Plus you don’t know what the player was thinking or feeling when he was playing, so your in-house testers or focus groups may be more valuable. And from an analytics perspective, parsing videos is a pain compared to just collecting the logs you need. As an example: how can an algorithm know when a player died by just watching a video?

    “keeping game state in-sync between servers and clients in a safe, fair, and accurate manner is no small feat. With cloud gaming, these issues evaporate because there are no clients anymore.”
    Well, from my game network programming experience, lag is what harms fairness. And from a network perspective, it does not matter if you stream video with SIP (as I guess Gaikai/Onlive are doing), receive JSON from an HTTP server, or small and bursty UDP packets from a traditional MMO server: the lag is still here. As you say, there is an advantage to stream: the player’s view of the game is (nearly) perfectly in sync with the server. But the roundtrip to the server has actually increased (because you need to add the roundtrip between streaming server and game server to it), so in practice, players lose responsiveness.

    So yeah, clouds make storms, but right now it’s really only hot air: in many cases, it’s not the most appropriate solution.

  3. @gentimouton I tried to approach the issue from a design, not a technical, perspective, so I am sure that there are still open questions about what is and is not possible.

    However, concerning having no actual client, I am talking about the advantages from the developer’s perspective, not necessarily the player’s. This change would be huge because client-server architecture (such a common system that it’s usually not worth talking about) is a big challenge and one of the reasons online games require such large teams. When smaller teams can make the same types of games as major studios, creativity flourishes.

  4. @gentimouton: I advise looking up nVidia’s recent news about their GeForce GRID platform (developed with Gaikai for their next gen servers). They’re basically using headless (slightly underclocked) GeForce GTX690s with double the RAM volume (with up to four 720p/30 feeds being rendered and encoded to H.264 on each graphics card)

  5. Soren, why the rose-tinted glasses?
    For such a lengthy article, I’d expected at least a little bit on the downsides of cloud gaming, for gamers and developers alike.

    What about cloud gaming preventing players accessing games when their connection is gone, or just slow? What about the problem of no more game ownership by players? What about privacy concerns? What about the problem it makes for modding? about centralizing the control of gaming even more, to the logical conclusion that less-popular games will simply disappear as thay’re not as profitable (see EA’s actions regarding game servers)?

  6. Pingback: Soren Johnson: “So far, the big question for cloud gaming is when will it be feasible, but ultimately, the more important question is what will it enable.” » Matchsticks for my Eyes

  7. Pingback: 翻訳記事:来るべき嵐 | スパ帝国

Leave a Reply

Your email address will not be published. Required fields are marked *