Video Game Bots: A history

Making video game bots awesome again

If you were playing a lot of F.P.S. games from the late nineties to early two thousands, you know bots were a feature that seemed almost ubiquitous. From 1999 to 2004 is when bots were at their most ubiquitous for pretty much any game with a significant multiplayer component. There was a two out of three chance that it would come served with bots.


The vast majority of N.P.C. (non-player-characters) that you encounter in games are heavily scripted and intended only to cope with a very narrow set of circumstances that are largely controlled by the developer. They really don’t understand what’s happening. They’re just taking prescribed actions in response to pre-determined triggers, which makes implementation much less daunting.

Bots on the other hand need to be able to play the game as closely as possible to how any other players would play. So they need to be able to use all of the abilities that are available to real players. They need to understand and react appropriately to any situation that might arise that requires a lot more work and even in the best case it’s going to be reasonably easy to tell which players are real and which aren’t. And multiplayer games are non-linear and unpredictable. They’re meant to have a vast possibility of outcomes by design. It’s naturally going to be a more complicated process making a bot who can keep the game interesting over hundreds of matches in the same level versus about who’s just meant to walk into your gunfire and die. One is meant to do its job and the other is meant to create lasting value.

Having a good bot is very important as a great way to introduce new players to a game or provide practice opportunities for experienced players, support great co-op experiences, and experiencing larger scale battles with just a couple human players. Perfect Dark may still retain the benchmark for out of the box features you could fill a map with customizable simulants with variables not just for their difficulty but also their rules their personality and their individual sets of clothing.

Battlefield may have been the boldest and the most necessary franchise where bots were needed.  Battlefield maps are way bigger than the arena shooters mentioned previously and you have players regularly switching between F.P.S. and vehicular modes of gameplay, which means more different types of AI have to be made to play effectively over greater distances. Which means more programming more bug testing and more waypoint placement. It couldn’t have been easy but they did it.

Otherwise you’d need anywhere between sixteen and sixty four players to get an accurate snapshot of how these games were meant to be played and the further in history you go the less likely that is to happen.

Then, Halo proved that you could push out a big multiplayer shooter without bots and still push millions of sales just fine. From 2005 onwards bots were and have remained relatively dismissed. Nowadays pretty much any game with a big multiplayer component is not likely to have a true bot match mode. Even in bots that exist in newer games, the AI follows different rules than human players it’s not a snapshot of how the game was meant to be played and neither are the battlefield campaigns.

Why is no one making bots now?

So, where did all the bots go? And why should you care? You won’t find a whole lot of angry editorials or developer post-mortems about the good old days of bots and how popular they used to be. I’ve mostly been interested in playing with bots for the purposes of historical preservation – kind of demoing the game and seeing how it was meant to be played on a platform that maybe it wasn’t meant for or a time in which it wasn’t meant for.

It’s hard enough to get P.C. games or console games from twenty years ago working today. So imagine how much harder that will be two hundred years from now. This is why I care about bots.

One of the greatest challenges facing people of the future is going to be how to preserve digital media build for today’s machines. We’re living in a kind of precarious time where an incredible amount of human knowledge can be found with a quick google search but it’s always slowly going away and slowly becoming obsolete.

Even after getting an old game up and running again it’s kind of a wasted effort if the game was focused on online multiplayer. Bots, though, give us a snapshot of the past – they let us see how games worked in their heyday. They preserve gaming history by acting out a gameplay demo no matter the time or place.

Here are some potential reasons that bots are no longer meant to interact with players in video games, but instead, interact on chat apps and social networks. The conspiratorial answer would speculate about marketers pushing customers to online services, but most blame much less sinister factors.


So here’s my conclusion: the strenuous labor of placing pathfinding nodes versus the unreliable results of unscripted pathfinding, plus the high cost of programmers and the high demands of now skilled multiplayer gamers means that programming bots is simply hard work (unless they’re uBots, of course). Meanwhile it’s the future: we’re always on broadband internet connections and the higher pools of real human players that can jump into any little cheap steam release, plus the short term irrelevance of historical preservation via bots means that programming bots is no longer financially worth the cost.

There’s a lot of small multiplayer releases they could use bots – and there’s a lot of big multiplayer releases that could use this feature. And if the beautiful intricacies of a good bot match are destined to be forgotten with time, the future of bots will be least a little disappointing.