Last year I wrote a post about the effectiveness of making a video game with the intent of inducing EA ideas naturally through gameplay. The game got released on Steam and received positive impressions, so I wanted to follow up with the results.
Although I was originally quite optimistic about using games (and other forms of art) for EA, my current thinking has changed. First of all, the original estimates had various flaws:
1) Combining confidence intervals also increases the uncertainty of the variable. Thus none of the calculations were actually conclusive.
2) I underestimated the amount of effort/luck needed for marketing, and overestimated the expected number of players.
3) Although the game received praise for its depth, in 10 follow-up interviews none of the reviewers reported having changed their behavior or thinking as a result of playing it.
Since only around a thousand players experienced the main part of the game, having spent a whole year on it seems inefficient. Additional effort also seems unlikely to greatly improve the cost-effectiveness, even though the playerbase still has room to expand.
Given this I'm now less confident about whether game development can be reasonably pursued from an EA perspective. The effects don't seem tractable, it's very difficult to know what will be meaningful during development, there's loads of work that's not relevant to the intended message, and any larger influence requires a disproportionately lucky hit in the market.
As far as I know, it also seems that the people behind https://www.effectivegivingquest.org/ and https://www.twinearth.com/ have experienced similar problems, more or less abandoning their EA+gaming projects. (Correction in the comments: EGQ closed for unrelated reasons.)
That being said, I do still think the medium has lots of unexplored potential, it just seems very difficult for game developers to utilize. My guess is that people in lead design, director, and producer roles at large studios seem much more likely to be able to induce relevant insights for (a large number of) players. In comparison, spending lots of time and money for an indie game just to temporarily influence a handful of players doesn't seem like a very effective endeavor. Writing a piece of text would be a much leaner method of delivering a message.
Personally, I made the decision to focus more on AI alignment moving forward. It seems more immediate, more important, more tractable, has much higher marginal impact, and could also benefit from shifts in cultural norms. I'd like to recommend looking into it if you're at all interested.
Have you considered games of smaller scope which have more virality chance?
Not quite a game but an example ish: https://www.humanornot.ai/
One Chance https://www.newgrounds.com/portal/view/555181
I think there is opportunity to produce uncomfortable games. I'm imagining the famous pandemic flash game could be spun to be more EA related.
There could be some inspiration from Cold War era nuclear war movies where the message is clear just but showing the danger and result.
Personally I think there could be room in incremental games (e.g. A dark room) or social deception (with LLMs) displaying how powerful AI's current capabilities are even today.
Yes, it seems likely that an "interactive message" would have better value for the development cost. It might be worth trying in some cases. However, there's two main problems with this approach:
The upside is that any go... (read more)