Will's Blog

Prismata

About a week ago I played in the Prismata Reunion Tournament, where ~40 people came together for a game that's been effectively dead for the past 5 years. It was awesome!1

I discovered Prismata around 2016 while searching for cool browser games I could play at school2. The game was still in alpha but there was a playable demo using Adobe Flash, and I remember it being quite hard but very deep strategically. I played some more in 2018 when it came out on Steam in Early Access, and racked up ~250 hours on it before losing interest. (For context, most games I play end up in the 30-80 hour range. Prismata was a lot of fun.) Prismata died sometime around 2021 since it never reached a large enough playerbase to stay profitable, which is really sad because it's an amazing game that's quite different from anything else I've played.

About Prismata

The game is divided into two parts: one is a single-player campaign that introduces you to the game and contains "puzzles" to solve, which are pre-crafted scenarios where you have to figure out the right things to click in the right order to complete the mission. I've always been terrible at logic puzzles so I didn't spend a lot of time here. The real gem is the multiplayer experience, which has a ranked ladder, friendly matches, bot matches, and some wacky event modes that I never really played. For the rest of this post I'll be referring to "Prismata" as the multiplayer game.

The Gameplay

Prismata is an asymmetric, turn-based strategy "card" game with full information. There are 5 types of resources, two values called "attack" and "defense", and whoever fully destroys the opponent's field first wins.

There are 10 base "cards" that are available in every game, which include a bit of everything. Some of them produce resources every turn, some of them produce attack every turn, some of them provide defense, and some do a combination. Every game also includes some amount of extra cards, usually 5 or 8, which are randomly selected from a huge pool and range from slightly stronger than base to game-warping. All cards are available for both players to purchase at any point during the game (on their turn), though there is a limited supply of each card that is also player-independent.

Most turns taken follow this order:

  • Decide how to distribute opponent's attack from the previous turn among your defenders
  • Collect resources from economy units
  • Spend resources on buying units
  • Decide which units to attack with or to use their click ability
  • End turn

Because player 1 goes first, they start with a slight economic disadvantage. This means that for some games, one player will have a "smoother" line to build a certain economy, but in practice this advantage only matters in very high-level play. Prismata is otherwise a very "fair" game, since both players know what units are buyable from the beginning and can see everything the other player does. This can lead to very interesting midgames where one player doesn't want to commit to a certain build pattern too early, since the other player will immediately pivot to counter it after.

The Learning Curve in Prismata

There are quite a few level of bots in Prismata, and they're useful because the game has a lot of complexity. One of the easier bots, Adept Bot, handily beats most beginners, and I think this early wall unfortunately turns a lot of people off this game. The magic happens when you become more familiar with the mechanics and figuring out build patterns until you find Adept Bot actually plays very poorly, and even the strongest bot, Master Bot, is fairly weak. 3 However, this requires a good amount of upfront effort and more thinking than a casual game.4

Why is there this huge jump? Prismata is a very "snowball"-y game. Every mistake or suboptimal decision you make compounds over time, assuming your opponent doesn't make similar mistakes. In a game like chess, you can be behind a piece in the opening and maintain that deficit over time. But in Prismata, the difference grows naturally every turn until you're staring a giant 24 attack in the face while only producing 6 attack yourself. There's also no "hail mary" or comeback mechanic that will save a doomed board: in fact, the developers have mentioned that they designed the game so players in a losing position lose even faster. Most games end in resignation as soon as one player realizes they're too far behind.

Sadly, while people like to complain about RNG, it's also a bit unfun to be thoroughly crushed due to having skill issue. Once you're good enough to beat Master Bot regularly, you just need to play other people, and in 2019 that was my first experience with ladder anxiety. That was pretty uncomfortable, especially since I knew every loss was 100% my fault. Still, I enjoyed it enough to play it practically every day; I have vivid memories of being on vacation in Europe and still finding time to play Prismata.

The Future of Prismata

Despite the great turnout for the tournament, the game is still very dead. You can get a ranked game if you sit in queue long enough, but almost everyone that still plays is extremely good. This leads to the issue of "few weak players" -> "any weak players that play get destroyed by veterans" -> "weak players don't want to keep playing". I still occasionally play against Master Bot for the daily rewards, though.

I'm happy that the creators, Lunarch Studios, have had some success with their other games, but I don't think they have any plans to return to Prismata at the moment. In fact, Prismata's single-player campaign is actually unfinished! But for now I'm happy enough that they've kept the servers up so I can revisit this amazing game every once in a while.


  1. We played 10 rounds of Swiss before a top 8 knockout round. I seeded near the bottom and lost a couple games near the beginning, so I ended up playing some of the weaker players and ended 5-5. That's honestly way beyond what I expected! One of the games I played was streamed on Twitch by my opponent as well: I'm at 3:18:00. ↩

  2. This is also how I discovered Rhythm Doctor, Fallen London, Die2Nite, no-one has to die, Gods Will Be Watching and Exit/Corners (among others), which are all games that ended up having quite an impact on my life. ↩

  3. I have around a 90% winrate against it, and I'm probably around top 200-250 in the game. It's possible that with the advent of LLMs that a stronger bot could be coded, but Master Bot has been the strongest available for a long time. ↩

  4. If you've read my post on Hanabi, you might be seeing a trend here. ↩

Thoughts? Leave a comment