TF2 was the first online game I really played full-time hardcore srs-bzns style. Before TF2 entered my life, I’d occasionally get to try games on my friend’s N64s and I had vague memories of Oregon Trail and Humongous Entertainment games growing up, but I’d been largely sheltered from the gaming world due to my parents thinking that games rotted your brain. So once I left for college and experienced newfound freedom, The Orange Box seemed a logical first purchase, since I’d also played Portal when it came out for free in 2007 and (somehow) passed my parent’s “nonviolent and educational” criteria.

Every game in the Orange Box was awesome, but since TF2 was only one designed to be played infinitely, it obviously captured the lion’s share of my attention. And hoo boy did I become a fanboy. As my undeveloped brain slowly experienced various nuances of game design at the same time as I started making my own games, I remember thinking that TF2 was literally perfect at a core level, every little nuance handcrafted by gods among men who’d made the absolute perfect choice in every case.

And for the most part, I was right. TF2 is a very well-designed game, after all, appealing to many different facets of players and gamers. But now, looking back, I can think of several design choices my younger self thought was a brilliant stroke of genius, but I now see as something that probably hurt the game at a fundamental level.

1. Specialists and Generalists. When I first learned about how certain classes were designed to be played full-time and others should only be used in moderation, I thought this was a nice touch. It appealed both to players who like to mix their played class up frequently and those who want to do the same thing over and over forever. It also encouraged variety and differentiation among class.

Nowadays, however, I’m starting to think that this only serves to unfairly give certain classes a legup in the metagame. Either every class should be a generalist or (probably better) every class be a specialist reliant on his team to do his job well. Obviously unlocks make this difficult, as it’s far easier to design an unlock that turns a class into more of a generalist simply through giving the class more options,

There’s also the concept of stacking. My younger self figured it was impossible to design a game where stacking could be countered, but I hadn’t played Team Fortress Classic at the time, and as I think about that game, I’m realizing that it actually has no trouble with stacking, because a stacked team always performs more poorly than an unstacked one. If one team attempts to Medic rush the flagroom, they’ll find that their lessened defense will make life too easy for the scouts and medics on the enemy team. Likewise, a team that stacks defense will find that they don’t have the offense to ever capture the flag themselves, and the offense-focused game gives the initiative to the team that attacks more. A rush of any of the nine classes in TFC just opens up a glaring weakness for a balanced team to exploit, unlike in competitive TF2 where class limits must be in place to prevent uncounterable “stacks” of two Medics or two Demomen, much less even more that.

This distinction is even more interesting considering how much less balanced the TFC classes were among themselves. Certain classes were clearly just superior to others, like Scout vs. Medic or Pyro vs. anything else. I think the balance comes from its more dedicated dichotomy between Offense classes and Defensive classes. The concepts of “offensive class” and “defensive class” are way more integral to TFC, even though TF2 is the one that tried to define them. In TFC, the offensive classes (Scout, Medic and Spy) are powerful at their job but almost useless on defense, with the defensive classes (Heavy, Sniper, Engineer, Demoman) likewise far more powerful in their chosen role. The only outliers are Soldier, the only class truly capable of switching roles on a dime without a loss in efficiency, and Pyro who just flat sucks at everything. In other words, the classes are far more specialized. Even Pyro, who just decided to pick a crappy specialization and then couldn’t even adequately do that.

2. The shotgun. By which I mean that four of nine classes get the exact same one. My younger self loved when a set of options has one obvious outlier; in this case, I thought it was really unique that a single weapon made a swath of appearances across the spectrum in defiance of the game’s rigid dedication to class differentiation. (For this same reason I paradoxically didn’t like the pistol, for infringing on shotgun’s status somewhat.)

But now I’m not so sure Valve wasn’t just being lazy and sticking the shotgun whenever they couldn’t think of a better secondary. The only class where the shotgun was a clear candidate for use is Soldier, since the airshot has been something of a signature move ever since TFC days. The other three classes are of course capable of using a shotgun, but it wasn’t so integral that it couldn’t have been replaced with something else to give more of an appearance of variety. To the same extent the pistol and identical melee weapons hit this point as well. That’s not to say I don’t love the shotgun, but when one of TF2’s huge selling points is how different the nine classes are, I don’t think a ubiquitous weapon is helping that point.

3.Constant balance tweaks.  This one may seem odd, because I never really consciously thought about the fact that Valve is always releasing little tweaks to the classes. Of course that’s a good thing, right? Get the bugs out of the way, fix anything that doesn’t match Valve’s vision of the metagame and keep everything running smoothly.

But it wasn’t until I became part of a team of designers creating a Street Fighter-style fighting game and the lead designer was emphatic that we not do this that I learned there was merit to doing the opposite. He insisted that while we can playtest the crap out of the game until the moment of release, we will  NOT under any circumstances release balance patches after it comes out because at that point, the metagame should be out of our hands. If a character is controlled in a way we weren’t expecting, or if it turns out we grossly miscalculated the effectiveness of certain moves, we can’t touch them because the community needs to be able to rely on what they’re being given.

While Valve probably shouldn’t have handled things to quite the draconic degree that guy wanted, it is true that constant tweaks make it impossible for players to rely on new strategies. If a player discovers a cool new way to play their favorite class, it can be quite disheartening to see a future patch unsubtly nix the new playstyle. This is where a lot of outrage regarding certain nerfs and buffs come from. I think Valve could have been a little less balance-crazy in constantly changing things for years on end just because they wanted balance to go exactly the way they envisioned it. This resulted in a lot of baffling and backtracking and weapons with widely variable equip rates as they get their stats warped into new dimensions of power or suck. I think it would have been better if Valve had tried to limit themselves only to problems that were actually breaking the game and not hundreds of little ones like “removed 6 stickies from the Sticky Jumper’s clip years after the weapon came out.” Because I highly doubt that was a required patch to the current metagame.

A better source of addressing tiny balance problems, like the system we used in the design of the aforementioned fighting game, is through releasing new content that addresses the problems with the old content via introducing new counters and playstyles. It’s far more organic for a player to accept that a new character changes the metagame than a balance patch just crashes the old characters and you can no longer play them the way you used to.

The thing is, balance finds a way. Most metagames are fluid by nature. In the beginning, certain elements of a game (be them weapons, characters, moves, etc.) are quickly determined to be overpowered, and everyone uses them. This leads to players choosing whatever available counters they’ve got, and the metagame shifts somewhat. Occasionally, there’s a notable breakthrough when a previously-unnoticed quirk is found to work well in this new environment, either due to metagame shifts or just a player utilizing it in a new way, and the metagame shifts again. Developers often don’t need to step in unless the metagame is very badly broken in a way severely hurting the entire community, and I don’t think most of Valve’s changes qualify on that level.

Again, I feel the need to close with a reminder that  love TF2 and almost everything about it, especially on its core level. These are just aspects of its basic design that I think are less than stellar.