I didn't go into this with anything in mind, and while I often comment on my posts, wanted to avoid doing so until other people had taken a shot.

I think I'd probably go with at least one game that can be drastically modded or otherwise extended, because I am pretty sure that no matter the game, with only three games for five years, I'm going to run dry on gameplay prior to the term on the island expiring.

So I think my three would be:

• Cataclysm: Dark Days Ahead. On its own, it's a roguelike with a procedurally-generated world that I have kept coming back to for years, so there's already a lot of replayability there. It has very complex mechanics compared to most roguelikes, has been extended a lot. There aren't actually all that many mods, but that's because most people just add their mods into the core game -- the only reason something becomes a mod is if the devs feel that it'd drastically alter the balance or structure of the game (no total remakes), make it too large (e.g. there's only one small soundpack included with the base game), or if they have some other kind of issue (e.g. the modder hasn't been following copyright). So the base game has...a lot of stuff. I've done work on the code before, have the knowledge to do so, and I feel like that'd be a game that I'd be willing to add features to over five years, has a fair bit of potential there. !catadda@sh.itjust.works, though it's pretty inactive.

• Skyrim, for the existing mod content. Of people who have responded so far, this was a popular choice. I think that of Bethesda's games, in 2024, this is probably the game with the most-extensive collection of mods, though maybe down the line Starfield or something else will become bigger. There are very few games that have such an extensive library of mods and can be played in as many ways. I don't, in 2024, have a lot of the expertise necessary to do most of the interesting modding of it (e.g. texturing, audio, and animation), though I suppose I could pick it up. I regret not having in-game building that Bethesda's later games support -- I feel like having a sandbox-style game really would help longevity of play -- but as of 2024, all of Bethesda's later titles range from being somewhat-more-limited in mods (Fallout 4) to vastly-more-limited (Fallout 76). Skyrim also isn't the prettiest of the bunch -- or, even, my favorite to play not-on-a-desert-island -- but I feel like the importance of graphics and audio quality falls off a lot over time if you keep playing a game. The extent of the mods let the game fill multiple genres: the game can be played as an RPG (with modders drastically extending or even remaking that aspect) , a survival game with difficulty that can be ramped way up, a sexy adult game, a territorial conquest game, you name it. I think that the sheer amount of hand-crafted content that modders have put out will probably last for a long, long time. !skyrim@lemmy.world and !skyrimmods@lemmy.world.

• Minetest. I wanted a sandbox game that has few limits on how far a given run can expand, think that that probably has the most potential to remain interesting. Minecraft has far more mod content created by others, would be better to play, but I can patch the core Minetest game (and have in the past), and I think that for me, that tradeoff would give Minetest the edge. I also think that it'd be possible to, without adding a lot of complexity from a mod standpoint, make something that would be fun to play oneself -- I think that most Skyrim mods aren't really fun to play for their creators, because they're about experiencing handcrafted content, and if you've created it, I think that it's hard to enjoy yourself. But games with a procedurally-generated or evolving world...you can create the levers and still have fun pulling them.

Some things I considered but didn't do, or things that other people have suggested and, upon consideration, I don't think that I would do

• Some sort of conventional (non-video) game that it's demonstrated that people can spend many years on becoming expert in, like chess or Go, and taking one of the video games that implements it, has an AI to play against. I can believe that it's possible to become deeply immersed in those games for many years because, well, there's an established track record of people doing so. They probably aren't the worst choice, but I don't, as things stand, really enjoy any of those enough to want to be doing them for that period of time.

• Factorio was a very-popular choice, and someone mentioned Satisfactory. I've played it, like it, and I agree that you can spend a lot of time building out on a given run; there aren't a lot of limits on what you can do. But my thinking is that the core gameplay loop just isn't that complicated. And I feel like Minecraft-type games have more potential for very large and sophisticated creations.

• Games with a lot of fancy DLC. Of the responses so far, these were not a popular choice, even though I totally removed cost from the equation, which I'd kind of expected might favor them. I suppose that the real story here is that the DLC that any one developer has put out, even if it's a lot, kind of pales in comparison to what modders have produced...especially in terms of gameplay. Most developers don't radically expand how a game plays in DLC, but a lot of modders have in their mods.

• A solitaire engine that has the rules of many games implemented, something like PySol; I don't feel like that breaks the "only one game" rule, as I think that most people consider a solitaire engine to be a single game and probably the majority of solitaire engines out there implement multiple solitaire games. I've spent a lot of time playing solitaire games, even if I don't get wildly excited about them, and particularly like Eight Off, which is a bit like an easier Freecell. But it doesn't tick a lot of the other boxes that I feel would be important, like being interesting to mod -- I don't really want to create new solitaire games -- or being a sandbox game.

• Kerbal Space Program. I thought about this one. That was picked by some other folks here, and I think that it's a good choice, as it's got a sandbox aspect, a lot of mods, has a lot of long, hard stuff to accomplish in-game, has successfully held my interest for long periods in the past. Just didn't edge out the ones that I did pick.

• Terraria or Starbound. A lot of mod content, and they benefit from the sandbox aspect. And I think that they're good games; I personally favor Starbound, as Terraria is a bit more story-oriented and I think that that aspect would lose value when the game is played a great deal. I also think that there's more mod content for Starbound out there. But they lack the ability to create much by way of automated environments, the sort of way the Minecraft genre has, and I think that that limits a lot of what one can do to creating large projects that are cosmetically-interesting. That's okay too, but the ability to create large projects that do things, I think, has more potential.

• City-builders like Cities: Skylines. That was chosen by some people here, and I certainly don't think that it's the worst choice. Problem is that I think that a lot of those are about experiencing the content, and that you're gonna run out of that before all that long, and that the core gameplay doesn't get wildly extended via mods or DLC. I think that there's a point where one's pretty resoundingly beaten the game and that it's less-amenable to pure-sandbox stuff, like creating fantasy cities, than something like a Minecraft-type game is.

• Dwarf Fortress. A few respondents so far have chosen it. It's not a bad choice. But Cataclysm: Dark Days Ahead ticks a lot of the same boxes and for me, edges it out (even if Dwarf Fortress is mostly about simulating a colony of dwarves with some limited single-character-oriented content that was added and Cataclysm: Dark Days Ahead is mostly about single-character-oriented-content that added some limited colony-oriented stuff). And I really want at least one game that I can personally hack on, because that opens up a lot of potential to be interesting, and DF isn't open-source.

• Oxygen Not Included and Rimworld. I love both these games. Also popular choices, and a lot of gameplay potential. Not the worst choice. The mechanics are simpler than something like Dwarf Fortress, though, which I think would be a drawback if one commits a lot of time to them, and I don't know how easy it'd be to extend them in interesting ways. From memory, the existing mod libraries mostly don't radically change the way that the game is played. Like, they'll add more animals, more factions with different appearance and slightly different stats, something like that. I think that having large mod libraries that make the game playable in significantly-different ways would tend to add more to replayability.

[continued in child]