Streaming, Game Instrumentation, and Better Experiences

Happy New Year!

I’ve been watching a bit of video game streaming of late, and one thing that’s struck me is that most games aren’t instrumented to accommodate stream integration. I couldn’t find much information on the subject, so I thought I’d scratch out a few thoughts.

Streamers may want to track in-game deaths. That should be trivial with an API (and it may already be possible with game mods). Games should absolutely provide some kind of event stream that can easily be integrated into streamers’ on-screen displays. There are a wide variety of possibilities this opens up, including better multi-stream races (where the programmatic reporting of milestones can be plotted on a simple race chart) to better and automated tagging of stream clips (e.g., automatically linking to significant in-game events).

The Steam platform has added some game-tracking for their social component, so that you can see what your friends are playing with a little more detail, but that’s only a baby step. Valve’s own games also feature statistics, and with the advent of GDPR customers can see more of that data than ever, but there’s a lack of tools to connect that sort of data into something that would improve game streaming.

What else? How about viewer experience? The visual environment of the player can and should diverge from the viewer in some ways (with the viewer still having the choice to watch the game footage or the enhanced version). For example, Counter-Strike: Global Offensive has a spectator mode that shows all players, shows grenade ballistic arcs, etc.

At some point, it may even go the other direction, with viewers being able to influence the experience of the streamer by causing enemies to spawn or such. Watching a charity stream earlier in the year, they played some Jackbox Party Pack 5 which lets viewers interact through a website/per-game password combination (rather than directly from the stream chat) in order to avoid the streamers seeing the viewers’ answers. There are also a few games like “Marbles on Stream”, which let viewers “play” by assigning their name to marbles in a physics simulation/marble racing game and see whose marble wins.

The interaction model may have to change a little, such as having streamer-blind chats for the purpose of letting viewers have more control without “stream sniping” (when someone can gain advantage by watching a stream or chat).

Some work on stream-and-chat interactions have already been done with the famous Twitch Plays Pokemon and the like. This seems like very fertile soil, and it seems reasonable to expect that game makers will start to implement things to let it develop and mature.

The Valve Mod Marketplace Fiasco ARG

So I’d written a piece that painted the paid mod controversy as a new alternate-reality game by Valve. But since the whole thing is on hiatus, I guess it won’t work.

What can be said, instead?

I think Valve is right about the inevitability of paid mods and having a more fluid system for moving works online from free to paid. They just didn’t have a very successful rollout. Part of that was the 75% rake between Valve (30%) and Bethesda (45%), leaving the mod maker with the smallest share (25%). Sure, the money spends for a mod maker that would otherwise get none, but it rubs the buyer the wrong way.

There’s a lesson in that. If other semi-predatory industries like the music industry had a more prominent display of how little the artists get out of your $15 album purchase, it could shake things up a lot. And that goes for other industries like farming, clothing manufacturing, and so on. If people know that workers are getting screwed, they’ll at least make a stink. If they can just ignore it, because it’s not in their face, they’ll tend to ignore it.

There were issues with misappropriation of others’ mods. Valve will have a hard time working out a perfect model for derivatives and dependencies on the legal side of the issue. But they can at least push for better technological integration of mod dependencies in games.

And Valve is right to glimpse a future where games themselves might be seen as a greater-than-the-sum-of-their-parts assemblage of mods. Something like a patchwork quilt that you play on a computer. That future will come to pass in time. It won’t be exclusive, other non-mod-based games will exist. But it will live alongside those games, both feeding off them and feeding into them.

In the meantime, it appears that the factions I’d described in the hypothetical ARG seem to be here to stay. We will probably see mods that will license themselves only for use with free mods, for example. While others will say they’re happy to be used by paid mods.

But paid mods do give modders an incentive to work and a mechanism to buy work from others to make their own mods better. If you’re doing free mods exclusively, you might want to get some better textures or models, but have to take what’s free. If you sell the mod, however, you can afford to hire professionals to augment your abilities (e.g., if you’re writing code, you can pay other professionals can do the art) make that mod a bit better for customers.

The other thing this whole incident reminds us of is that we will undoubtedly see other monetizations come forward. You might earn gametime or rewards in future games by helping new players out (as a guide would through a dangerous environment in the real world), for example. Or you might earn real money for doing so, as some already do by streaming their gameplay.

The nature of gaming is so digital that it provides a key ground to try things that might not fly in other industries, and although Valve didn’t get it right the first time, I hope they keep working on it.

Save Games in Gaming

One of the features most story-based games have is the ability to save your progress. Although some game companies are more focused on multiplayer games, singleplayer isn’t going anywhere, and multiplayer may include some singleplayer components in the future.

But if you’ve played a handful of games with a save game feature, you probably noticed they all work a little bit differently. And those differences may actually impact your play style.

Up-to-you Save Games

The up-to-you style is where the game simply has a feature and you can save whenever. It doesn’t save for you, and often you can name the save games. This style seems adequate, but in my experience I would only use one or two out of dozens of available save slots.

Usually I would stick to one slot, and then if things got hairy or I thought I might want to keep the other as a known-good save, I would add another. After progressing far enough, I might repeat that process since the known-good slot had aged too much.

Up-to-us Save Games

This is the opposite extreme. The game saves when it wants, and you have no control over it. In this sort of game, I would tend to be increasingly cautious as I moved farther from the last save. I would know that if I failed a mission or died, I would have to go all the way back. That inhibits risk-taking, which is not what games are about.

Games can encourage risks and teach us about when risk is appropriate. But if the game administrivia function itself discourages risk, that limits the game experience.


A number of hybrids exist. You get automatic checkpoints, which protects you from having a lone save game in a situation where you can’t progress, but you can add your own on top of it, giving you some say, too. This model seems the best of those currently available.

The main pain point with hybrids is that you might tend to accumulate save games that aren’t useful in the future, but if the save data is small and disk space is abundant, it shouldn’t really matter.

The Sandbox Model

Some games aren’t normal story-driven games, but sandbox or open-world games. In these what’s generally saved is the state of your playthrough: your inventory and completed missions, plus some sort of general location. This seems to be a pretty strong model for this type of game, but it doesn’t necessarily translate to more linear games.

Early games didn’t have saving available, though some used elaborate codes, as consoles lacked storage. They also had limited lives, which meant you could have to start the whole game over if you died. We’ve come a long way since those days, but I’m hopeful that future games will have save options that are even more powerful and reduce the friction of playing the game.

Half-Life 3 Speculation

Gabe Newell, head of Valve, recently gave an interview (SoundCloud: GameSlice: “#1: Gabe Newell and Erik Johnson from Valve”) where he spoke briefly about the possibility (or lack thereof) of Half-Life 3. This isn’t the first time he’s said things along the lines of, ‘we want to do it, but we don’t know how to do it with what we know now.’

Valve started out in the single-player world of games. The first Half-Life had multiplayer, but it was deathmatch only. Where the multiplayer code shined was in mods like Counter-Strike. Since then Valve has gone on to do more and more multiplayer and a lot of different market strategies with that.

It’s like seeing what you can do with the tools in a buddy’s woodshop with fancy powertools and then going back to a pocket knife and a stick. They don’t want to make another single-player linear game like Half-Life was, and they don’t know how to build into that universe in a multiplayer way (or if they do, they’ve not said so).

But they have a lot of data:

Team Fortress 2’s Mann versus Machine mode

They have some idea of how cooperative gameplay against an AI opponent can work. That’s not to say a potential multiplayer HL3 would look anything like MvM, but it is data they’d consider in building it.

Dota 2

They have some idea how cooperative and competitive go together, including AI friends and foes. All of these are present in Dota 2. It’s not clear if people would want to play as the Combine in a new HL3 game, but the possibility exists they would and could.

Others’ games

Valve also learns a lot from other games. Games like Borderlands 2 that feature cooperative play might give one possibility for a HL3 that isn’t all player characters, but where the core of heroes are people. Whether Valve would attempt a mission-driven game with maps like Borderlands 2 is an open question.

Valve also has content that never saw the light of day. Things like the commander class for earlier iterations of Team Fortress 2, which would have made it a partial RTS game might be something they look at and revamp for HL3 or it might not.

Ultimately, what HL3 will be isn’t as important as what it will contain:

  1. Freeman
  2. Scientists and allies
  3. Hostile aliens (headcrabs and zombies, plus others) and hostile humanoids (Combine or military)
  4. Gman

That’s the essence of Half-Life. The main challenge for multiplayer HL3 is that everyone wants to be Freeman. That was part of the appeal of the series, that you’re this lowly scientist that’s saving the world (and yourself). To suddenly break away from that and say “We’re all Robert Paulson” is a little cheap, but probably a necessity of a multiplayer Half-Life game.

It can be done, and done successfully. The message of mass movements is that everyone can carry part of the load, and that’s a very powerful message. But you still have the hanging string of Freeman to deal with. Is he dead? Moved up to management? Missing? Selling vacuums door-to-door?

Rise and shine, Mister Freeman. Your vacuum route awaits.

Interchangeable Video Game Parts

Often when I do play video games it will be on servers that use custom maps/worlds. These vary greatly in quality, from top notch to bottom of the barrel, but mostly the people loyal to the server do their own quality control and the better the server, the better the average quality is.

But the experience is still limited. Most of the maps are not open source. The assets generally cannot be migrated from game to game, particularly if the gaming engine is separate.

There’s a great opportunity for assets to be reused, improved, and evolved over time, to the point where whole, seamless worlds of assets could be constructed and used by many games.

I’m not talking about having a consistent world across multiple games, though that’s a possibility as well, but just sharing the world.

This already happens on a limited basis. As said, custom maps are used on a variety of servers, some of which have their own custom game modes. Other maps have been released by their authors for multiple games or multiple versions of the same game. Or multiple versions of the map, with variations.

High quality maps don’t come easy, they require a lot of exacting work. And then testing. Reworking. But the idea that there should be an equal number of high quality maps as there should be high quality games, to me doesn’t add up. My understanding is that the act of creating the game world is one of the most laborious parts of producing a game. Even dud games may have beautifully architected levels.

It just seems like a big shame to waste these levels, by having them used once or twice and never again.

Sound effects for films have been reused as long as they’ve existed. There are some famous ones, and others that are not famous but still recognizable if you listen. Point being that reuse doesn’t necessarily detract from the entertainment.

There are tangible benefits to reuse beyond the fun of the games. They help level the playing field for game design a bit, where at least some of the code for handling the maps is free and open. They give more designers an opportunity, as they can market themselves to a wide, inclusive gaming community at once, rather than one small niche of a particular game at a time.

It may also help to keep old environments alive. I’ve not played the likes of Quake or Doom in years, but I still recall bits of their levels. While I doubt any risk of these games being lost forever, I also doubt they will be often adapted to new games except through a down-in-the-dirt recreation from the skybox in (or maybe from the spawn entities out?). It just seems like being able to pull up one of those worlds and stroll through could and should be as easy as something like Google Earth.