A couple of months ago I conclusively answered that age-old question: “what are The 458 Most Influential Games Of All Time?”
Along with a plucky band of about 130 of @retroremakes Twitter followers, I compiled a comprehensive, totally unbiased and statistically rigorous list of titles that have all left deep and lasting impacts on the landscape of gaming. The tweet that called for submissions to enter into the “all time” influences list asked for around five suggestions from each respondent, and resulted in 458 unique entries, mostly with only one or two votes each. That demonstrates just how diverse peoples’ gaming histories are, and by implication, the diversity of the history of gaming itself. (How this history is being eroded through a failure of curation is a deep and serious topic for another time.)
I proffered my own five top influences of course, and I was fairly certain of my choices. Then I started to wonder whether I could come up with a long-list for my own gaming history. So I set myself the task of listing the 458 games that had the greatest influence on me personally, in decreasing order of impact.
Like an idiot would do.
It turned out that just remembering 458 games I’ve played in the 30-odd years (christ) since getting a ZX Spectrum on my seventh birthday was a challenge in itself. I proceeded by platform, considering each computer or console that I’ve owned in turn, listing the games I had. Then I racked my brains for arcade games I played in Blackpool in the coin-op heyday of the 1980s. I included some board games and tabletop games, in common with the public’s listing, and even a couple of early LCD-screened hand-held games.
By the time I had 400 games on the list I was really struggling. I resorted to trawling through archives of nostalgia like World Of Spectrum, and the full Atari ST Medway Boys Compact Disk index, to jog my memory. These led to surprisingly few additions, but slowly, surely, the total crept up until I passed the magic number 458. In fact I was close to 500 when I finally stopped.
I threw them all into Excel and stared at my accomplishment, my magnum opus: a huge, diverse, unordered list of… oh.
After ten minutes of manually shuffling the list around it became obvious that arranging these games into a descending schedule of influence was going to be a problem. Sure, it would take ages, but I could just chip away at it gradually. But time wasn’t the biggest issue. My biggest doubt was over my own consistency.
I initially determined that, given a choice between just two games, I’d be able to decide which had had more of an influence on me. I wrote a little command-line tool that presented me with many such one-shot simple choices, one after another after another, and ordered the list based on my responses. But then I realised that It would be hard, if not impossible, to avoid cycles of undefined ordering: Game A beats game B, and game B beats game C, but then a while later I claim that game C beats game A. Given the diversity of my chosen games, from so many genres and eras, it would be impossible to manually avoid making such errors. I could encode this checking into the application itself, but it was getting more complex than I’d like.
And this raised another issue. Were my individual choices even repeatable? The reasons for these games having influence over me are often very personal. Changes in mood might lead to inconsistent cycles just as easily as simple confusion would. And if mood could cause such cycles to appear, isn’t it possible, even probable, that even for a given pairing I’d make a different choice at a different time, in a different state of mind? This cast doubt on the entire process.
So I settled on a different approach. I’d treat each game as a player in a meta-game, and give them an ELO rating like in a chess ladder.
The score would start at 1000 for every game on the list. My application would offer me simple choices as planned, A vs B, but rather than directly ordering the list based on a binary response, it would update the score for each game. And the way chess ratings work is that the change in score is dependent on the likelihood of the result.
Now I don’t really play chess any more. It has been an influential game in my life, but I’m not very good at it and find it frustrating. But imagine if I played Garry Kasparov at chess. I’d expect him to probably win, like, four out of every five games or something. Maybe that’s generous, I don’t know, but my point is that him winning is the expected result. So when he wins, his score won’t go up very much and conversely my score won’t go down very much. But, on the occasions when he loses (which would happen, I assure you) his score will take a significant hammering and, conversely, my score will jump up fast, as fast as I would jump up onto the board with my trousers around my ankles chanting “loser loser.”
You can read up on how the ELO scoring system works for yourself if you like: how the formulae are derived, the statistical justification, and its shortcomings in some of its less-than-rock-solid assumptions.
Suffice to say that with little modification I decided I could use it to order my influential games list. And it worked pretty well. I’ve been spending no more than 30 minutes at a time repeatedly answering simple binary questions: has Lemmings 2: The Tribes influenced my life more than Micro Machines? Did Vib Ribbon have a greater impact on me than F-29 Retaliator? It’s really quite meditative. I can make about a thousand choices during one of these sessions, and see how the list rearranges itself.
I was right to question the naive exhaustive approach, because in every thousand choices there were perhaps ten that I either changed my mind about shortly after answering, or that I knew are really impossible to call. These were the cases that would have led to inconsistent cycles. But because this was now a statistical process, with unbiased random pairings and implied information about previous results “baked” into the scoring system, I knew that it would be somewhat robust against small errors. It would converge on a sensible solution. At the end of each session I reported how much the list has changed. Once the average move in chart position fell below 1% of the list I removed the bottom half of the list as “done” and started refining the top half. I repeated the process until there were only a handful of games left, which I could order manually.
I’ve also realised that there may be games that have slipped my mind, that I need to patch into the list as we go. There may also be new games that come out that have immediate influence and need to be added. In fact this has already happened once with the excellent Nex Machina by Housemarque. Locating where a new entry should go in the list that’s already sorted is simple, but it does mean that the final list of games will probably be slightly longer than the original 458.
There are loads of very influential games missing from my selection. I never had consoles as a kid because my parents believed real computers were educational, so I never played certain games. There aren’t many popular multiplayer games either because I tend to prefer a good single-player experience.
There are also some unusually lousy games high up on the list, and conversely some excellent games down in the bottom half. This list is ordered by the impact a game had on me, and that impact may not have anything to do with the quality of the games themselves. A good example is Cauldron 2, which scores highly not because it was a great game, but because it was the most expensive game my parents had ever bought me at the time and it was atrocious. They were reluctant to spend their money on other full-price Spectrum games for a while after that, so it had a strong but negative influence on me.
Basically, if I can give you an anecdote about a game, for any reason, it’s likely to out-score its more middle-of-the-road peers. This is a personal list of my influences, which means you can’t tell me I’m wrong. Though I suspect you will.
So what’s next for the project? What will I do with this big stupid list of things? My intention is to reveal them a few at a time, as a long-term project. Initially I won’t be going into very much detail, hoping to rattle through the bottom half of the list fairly quickly, perhaps dwelling on a few interesting entries as the mood takes me. Later on in the project I’ll begin to explain my working a bit more thoroughly, go into the history of more of the entries, and discuss exactly what made these games so important. In fact some of the entries later in this series will inevitably be somewhat autobiographical, and probably quite emotional for me.
In doing so I hope that we’ll all learn something about how I came to be the kind of idiot that would do something like this.