It Failed
Artifact currently hovers at a max concurrent player count in the hundreds, an order of magnitude below Yu-Gi-Oh! Duel Links. Steam reviews are below 50% positive, with recent reviews "mostly negative". As I type this it has 139 Twitch viewers. (Note: These were the metrics when I began writing this blog months ago - they’re worse now)
Unlike games like Fallout 76 and Anthem, which endured rounds of public humiliation, Artifact has been a very quiet failure - but a failure nonetheless. After months of barely updating the game Valve announced that they would stop updating it entirely, instead focusing on a long-term rework.
My goal is to explain why this happened. Why a game in a popular genre, that appears competently made by a well-respected developer, landed with a quiet thud. The simple version of my answer: it’s just not well-designed.
Judging Games on Apparent Craft
I suspect some will reject this out of hand - much of the analysis I’ve seen online is “the game may have problems but you have to admit that it’s brilliantly designed!” The initial consensus was that design of the game is a strength and that it failed for vague reasons like “poor market fit.” (A trivial observation true for any product that underperforms) But I'd suggest something different: that Artifact looks like a well-designed game but isn't one.
Artifact is clearly a professionally produced product. It feels like a game made by smarties - people who did well on their SATs and went to Carnegie Mellon. It has that good product look - the closest a card game has come to being a prestigious 3rd person sad dad simulator. I think there’s a strong temptation, especially among game developers, to look at the game from afar and declare that it’s a brilliant design because it looks like the kind of game for which that should be true. It fits the profile.
For this blog to have value you must accept that a game can have high apparent production value, a strong pedigree, and superficial indicators of quality while ultimately being no great shakes. And for the remainder of this piece I'll do my best to convince you that that's the case here.
A Monetization Aside
This is a blog about design but it would be almost irresponsible to talk about Artifact’s failure without mentioning the monetization. If you’ve heard any criticism of Artifact it’s probably this one, so I don’t think I need to elaborate much. But to summarize Artifact includes four different monetization models: up front payment, paying for packs, a transaction fee on card sales, and a cost of entry for the big-boy modes. I once heard King of Fighters 95 described as one big fireball/uppercut experiment; Artifact feels like one big monetization experiment, a Frankenstein’s Monster of monetization strategies.
Some have argued that the monetization is fine or even generous. That it’s not hard to “go infinite” and play forever without paying, or that the total cost of ownership is low. Those people are wrong. How they are wrong could be a blog post in itself, but even if they’re right (which, to be clear, they aren’t) the fact that the game has such an apparently punitive monetization model is a huge problem. But not the problem. Plenty of people jumped every monetization hurdle and still quickly stopped playing the game.
Issue 1: It’s Hard to Learn and Conceptually Confusing
A frequent complaint from Artifact players is that they don’t understand why they won or lost and don’t grasp how to improve. What Artifact appears to be about differs dramatically from what it’s actually about, making intuitive strategies incorrect.
The Deceptive Unimportance of Long Term Value
A common theme in European-style board games, like Dominion or Ticket to Ride, is managing potential vs kinetic energy. In these games players choose between actions that build up future potential and actions that expend resources to make immediate progress. The high-level strategy is recognizing when to be the tortoise vs the hare - when to hang back and build up steam and when to beat other players in a sprint to the finish.
At first glance Artifact appears pressed from this mold. Prior to release the developers heavily stressed the “modification” system; modifications are long-term investments that permanently buff cards. The item-buying system in Artifact also looks like a variation on this theme, allowing players to buy cheap low-impact items or save up for higher-ticket more impactful ones. With 3 gold you can immediately invest in a lowly Traveler's Cloak that gives a hero +3 health, or save up 7 gold for a Fur Lined Mantle that gives a hero +8 health instead. The latter represents more power per-card and per-item-slot but is slower - the tortoise approach.
There’s just one problem: Artifact games are too short for the tortoise approach to be viable. Even in draft, which is a relatively slow format, games often end in 5-7 turns.
When a hero dies they sit out a turn before returning to the field. If they die to an upkeep kill (die before they can act) they effectively sit out for 2 turns. Modifications and high-cost equipment persist when a hero returns after dying, but by the time a modified or equipped hero dies and then returns the game is often nearly over.
As a result many modification cards are bad, and the ones that are good would probably still be good even if they granted only a non-permanent buff. It’s telling that one of the best modification cards, Time of Triumph, costs 8 mana and is used as a “finisher” - it can win the game immediately and often represents no long-term value at all.
Midrange items in Artifact are almost entirely useless. Saving up to buy an item on turn 4 that provides good long term value makes little sense when the game can end in 3 more turns. The specifics of item buying also work against midrange items. Items are presented to the player in random order, and to reshuffle your item deck you either wait a turn or buy the topdeck item and draw again. So the dominant strategy is either to load up exclusively on cheap items, or use mostly cheap items plus a few high-cost game-winning items that you can quickly cycle to.
Pre-release the developers stressed playing for value as a core concept, but the way games play out in practice value-oriented play is usually a trap.
Killing Heroes....Is Good?
Heroes are the centerpiece of Artifact, so you’d figure the game is largely about killing enemy heroes while keeping yours alive. But this too is wrong.
One of the quirks of MOBAs is that while the goal is to kill the enemy structures trying to do that as quickly as possible is often counterintuitively a bad strategy. Hard pushing a lane can leave you open to ganks while having minimal upside. While Artifact doesn’t import this exact design quirk it imports that same sort of quirk, where the obvious thing to do is often the wrong one.
In Artifact once a hero enters a lane their ability to move to another is very limited (especially in draft format) - consequently a key strategy is to strand an enemy hero in an irrelevant lane where they contribute nothing to the game. When an opponent's hero is stranded in such manner killing them can do your opponent a tremendous favor, allowing them to redeploy that hero on respawn into hotly contested lanes.
Imagine a Chess where killing the enemy Queen was often wrong - that would be weird. (Yes, I know gambits are a thing)
Because Artifact encourages periods of inaction followed by combo turns (more on that later) killing heroes at the "wrong" time often has little upside. If you kill an enemy Zeus on 6 mana they can't cast their powerful 7-mana spell on the next turn as they'll be dead. But if you kill an enemy Zeus on 5 mana they respawn in time to cast their 7-mana spell on curve, and can redeploy Zeus to a more advantageous lane. As such killing Zeus of turn 5 is often irrelevant or a waste of resources.
Artifact appears to be a clash-of-heroes game but killing enemy heroes is often just plain wrong.
It's Noisy
Artifact has many random factors and involves many micro-decisions, which makes evaluating the impact of those decisions difficult.
When you lose on turn 10 it could be because you made the wrong choice on turn 9, but it could also be because you made the wrong choice on turn 1. In a game like Hearthstone, where in some matches there are only a few reasonable plays each turn, it's possible to track back to an early turn and understand how the game could have unfolded differently. But in Artifact this sort of analysis is difficult due to the number of possible decisions, which is further compounded by large amounts of randomization obscuring the outcome of those decisions.
People often compare Artifact to Poker (especially when defending the RNG aspects) but there are two key differences: Poker is constrained enough to be mathematically well-understood, and Poker hands go by fast enough to build an intuitive understanding in a reasonable amount of time.
When you sit down to play Poker for the first time and luck out on an inside straight you might assume this is a winning strategy, but you’ll quickly be disabused of that notion after a few hours of play. It’s also simple enough to calculate or memorize the odds of common scenarios in games like Hold'Em.
In Artifact you may consistently be doing the wrong thing on turn 1 and as a result losing on turn 10, but you can’t play enough games to spot a pattern and it’s hard to evaluate the impact of plays in a sea of other plays and random events, either intuitively or mathematically.
The end result is, again, people losing without understanding why or how to improve. The game has no real ranking, no ladder, no replays and no meaningful stats, so if you want to track the performance of various cards or strategies you best bet is to fire up Excel.
Issue 2: It’s Not Fun
A common theory of game development is that there are two main ways to make a good game: include great stuff and/or remove bad stuff. If you’re a weirdo who thinks of game design only in liquid analogies this is the juice and oil of “juice it or lose it” and “oil it or spoil it”- “juice adds pleasure, oil removes pain.” (My own catchphrase, “balsamic vinegar it or rejigger it” has yet to catch on)
There are plenty of good games that focus on just one of these two things: games like Killer 7 and No More Heroes include fun inspired ideas with low polish, while Blizzard games remove pain points while having few novel elements.
Unfortunately Artifact fails to execute on either of these. When I say it’s not fun I mean something specific: there’s very little wow factor or inspiration, and rather than avoiding pain points the game often seems centered around them.
Heroes Are Bland
Heroes should be the centerpiece of the game but are often simple stat bundles.They have mostly underwhelming (or no) abilities, and what should be signature abilities are instead moved to separate cards. (When you include a hero in your deck you also include 3 copies of their signature spell)
One of the joys of card games is opening a new pack of cards and considering how to incorporate them into a deck. In Artifact you’d think that heroes serve as the “build around” cards - the linchpins of novel strategies. But hero design in Artifact is very conservative. How do you build around a hero that does 1 point of random damage a turn? Or around a hero that has reasonable stats and no abilities at all?
