Forums  /  Don't Starve  /  Category Changes and Update

Hey guys, Misuto here.

Presently, our games categories are mostly dead. Of the four categories, two have 0 submissions, and two are sheer luck based on world gen.

Let's look at each category first and discuss the flaws before I propose new categories.

Doorway to Adventure:
This is pure luck, start a world, run towards a forest, if you pass 30 seconds and didn't find the door restart. This is our first category, the one people see when they come the and it's the biggest joke of a category. Keeping this is fine, but definitely move it to the back of the list, this should not represent our game.

Best Ending:
This is also pure luck as you can see in my most recent run. While world generation tends to follow rules, there are two checks made during world generation and and if the thing set pieces weren't spawned in following the rules they tend to follow (near a road, not in mosaic biome, then more specific rules for each like not a swamp for box and potato thing), then the game just breaks the rules and spawns it anywhere there is space. If it fails to find space then it just deletes anything non critical to make space leading to world gens where a set piece can randomly be in the mosaic biome no where near a road.

So, while one world gen may require an unlucky amount of exploration that punishes you for knowing the general world generation rules, another world gen could have all the things along a single road. This run, as it is, is just a longer version of the luck sack that is doorway to adventure. This run is nothing but start a world, run on roads, hope for luck, reset at 10 minutes if you didn't win.

Beat Adventure Mode:
This is a more skilled version of best ending as you essentially have to do best ending 5 times in a row. Honestly, this is probably our best run... Yet it has no submissions, not even from me. It's a long run. Each adventure mode is made to have multiple locks in place designed to slow down and test the player.

So while 30 minutes is an amazing run on Best Ending, 2.5 hours would be virtually impossible given the required amount of luck for adventure mode. 5 hours even would definitely be a great run in this category.

That's a 5 hour, no break, luck heavy, endurance speedrun. Yes there are (much) longer speedruns that some people do, but it takes a lot of dedication and interest.

All Bosses:
This speedrun could be fun in theory, but honestly should just be a deleted category. Rushing Ancient Guardian does in fact take a lot of skill compared to the other 3 categories (and likely would use the fireplace glitch to get an easy kill). However, the other bosses are just crazy luck dependent.

Spider queen, Unless your world has the queen's gathering set piece, the range of time for a spider queen to spawn from a spider den is [22.5 - 45] in game days. A ridiculously huge range. One player could take 3 real life hours longer (not including time spent paused on the map) than another just for this boss to spawn. Also, a spider den's growth is paused when it's not loaded into memory. This is why on long game worlds you can still manage to find level 1 spider dens far away from your base. There's a potential bug where going to the caves/ruins will also pause this growth.

Treeguard, also a luck dependent spawn but not time dependent. Time spent waiting on the spider queen can be used to chop down stupid amounts of trees while staying near the den to insure it's still in memory and growing.

Deerclops, this is the worst offender of luck. Deerclops has a 66% chance to spawn in the first winter, and in the first winter, if it spawns, it will spawn after day 31.5. So, over 4 and a half hours into the speedrun, you have basically a coin flip of having to hard reset and lose everything. (A reset is 100% required at this point, cause it takes more time to try to spawn him the second time than it is to restart and roll that first chance 4.2 vs 4.6 hours.)

This is another endurance speedrun that depends on a coin flip literally hours in...

Ultimately, only one of our current categories (adventure mode) is even OK. Though sadly unpopular due to it's time commitment. So how can we improve this?

Suggestions in reply to follow...

Casper likes this. 

First, Best Ending should be opened up to all characters. My new run doesn't eat, doesn't fight, doesn't craft, doesn't come close to insane, and only gets hit one time (and that was avoidable). No character has any sort of advantage in Vanilla DS for that kind of run. This shouldn't be different categories, just maybe a little note on which character was used in a run and perhaps a filter to show runs for a specific character. This make no difference to WR attempts while also making the game more appealing to people who have a favorite character outside of Wilson. I also think that this should not be our first tab run given it's sheer luck nature.

Second, Unless we brainstorm out a better category, at the very least adventure mode should become our first tab category. This is the most skill dependent category we have and maybe pushing it into the spotlight would incentivize people to actually submit even one run.

Finally, some suggestions on alternative categories:


Hat Trick: Same as best ending, but escape 3 times in a row on the same file.


By having 3 successive runs, we mitigate a lot of luck without removing it. Same principle used in card game tournaments. More games in a single match helps mitigate the chance of a highly skilled player losing due to a poor draw as a poor draw is statistically improbable 3, 5, 7, etc times in a row.

Furthermore, this run would be shorter and thus less of an endurance slog than the current adventure mode (3 easier worlds vs 5 more difficult worlds).

As for rules, character choice would be a factor again as a sheer luck do nothing but walk world 3 times a row is statistically improbable. If true balance was desired it would have to be limited again, but I propose that this level of balance isn't even needed. This isn't a hyper competitive game and would likely have very few runners. If it did become competitive and was getting more submissions than splitting categories would be needed but would be messy. (Also the most consistent run for this would very likely be Wickerbottom as she can most easily create a diving rod but this is saving a few minutes on a run that very likely would be over 2 hours. World gen would be a bigger factor than Wickerbottom's minor advantage.) So I say, open it up to all characters. Also, no mods of course.


Seeded Adventure Mode. This one has been mentioned before, but it would require an entire special made mod to make happen. So this suggestions is more of a dream as I personally don't have the modding skills required to make it a possibility.


If we hosted an official mod, that gave all runners the same 5 worlds, this run becomes less about luck and more about memorization, routing, and skillful execution. This would essentially be an entirely new mode modded into the game. This would be similar to runs in subnautica; a world you can learn the layout of while still having some luck to keep it interesting.


Those are the best two that I can think of. One to mitigate luck, and the other is kind of a pipe dream.

While these runs are pretty much dead now, it's actually a good time to start making this game more inviting to runners. With the Hamlet DLC announced for single player, we'll see more people playing and watching Don't Starve soon. I feel the best way to do this is move our focus from sheer luck runs to skill based runs. Adventure mode and/or hat trick should become our focus in my opinion.

Let me know what you guys think below. Would you rather let DS stay as a luck-only game or see us push a more skill based category before the release of Hamlet stirs up more interest in our game?

Casper and knewster like this. 

I like where you are going with this, I particularly think that the Wilson-only rule should be abolished. I disagree with it, and want to (and will) use other NPCs. If my runs end up being not suitable for, well that stinks, but I am going to use the character whom I like. I think it's a shame that the majority of speedruns of this game that I have seen on Youtube are not on, and I suspect the arbitrary Wilson-only rule is one of the major reasons why the leaderboard is so sparse. Apart from that, having a particular seed used all the time seems like a great idea, but apart from sharing a save file (which could be done) I can't think of any way to do it. Manipulating rng is something else that would make this game interesting, hardrocker recently posted something about it, and trying to find a great seed and then replicating it would make this run more controllable. I'd like to ask if we want people that run this game to always save when they do a seed right at beginning, then consider sharing that save file via dropbox in the event that it turns into a good run. If we can get video of them hitting the start button, we might even be able to time it so that we get the same spawns even without the save file with great timing (that may also be TAS only, sadly).

Casper likes this. 

I like all these ideas. I don't speedrun much, though I do watch a fair number. I find Don't Starve runs borderline unwatchable because of the clouds. I'd personally like it if there was support to remove the clouds with a mod, so as to make the run more watchable.

I think character selection makes the run more interesting. I know the run is pretty short, but I think you could come up with some interesting strats based on what character you are using. For instance, I could see Wickerbottom using sleepytime stories to get to the areas guarded by the clockwork automatons. (Not that I think it would be faster, but you get the idea)

Either way, I'd definitely be interested in more categories. Maybe some sort of "item-get" runs? I can't quite think of what part of the tech tree would be the end-goal. For kill all bosses, maybe a particular world-gen could be used to make the game go quicker. Shorter seasons, more often boss spawns, different starting season, etc?

MisutoM likes this. 

I totally forgot to mention how the clouds turn off both viewers and runners. We need a mod to disable them or to add a rule to prevent zooming out. There's no way a serious runner would give up the tremendous advantage that is seeing more of the world unless it is expressly banned.

Even the people asking me to do this run on stream change their minds within a couple of runs of just staring at the clouds. So it's currently a run that isn't fun to run because it's pure luck and it's not even fun to watch for the people who are interested in seeing Don't Starve played quickly.

It would change the run so much if they opened it to all characters (since character differences give no advantage in one world) and banned scrolling out beyond the point to which the camera defaults.

Alas, unless redlof joins us in this discussion there's nothing we can do?

Casper likes this. 

I thought maybe this was one of those games that had its own, more popular, fan site outside of the SRC community, but it looks like that's not the case. Is there a discord server for this game? I'd be interested in jumping in. I love this game, but I've never considered speedrunning it until recently.

The hat trick is a pretty neat idea. I think it's a good compromise to balance the luck. I didn't know there were rules for the thing spawns, but if there are, players shouldn't be punished for having more knowledge of the game.

I would be most interested in doing Adventure Mode runs, but disagree with requiring showing the world preset for them I don't remember if I saw the requirement in the thread above or on the site itself, but I don't think the presets affect Adventure Mode maps and it would just be a hassle to restart the map and find Maxwell's door every single time just to show the list of presets.

A note about the layout: because there are so few runs so far, I think that the board would look a bit cleaner if "character" was just another input option like time or console. It's how the SSBM board handles all the characters, and unless we eventually find out that one character is much more broken than the others, I think it makes sense to do it that way for this game too. At least for now. See here:


How about a 100% category where you would need to craft every item in the game ? I think something like that would make a speedrun of this game a lot more interesting.

oddtom likes this. 

So I'm a bit tickled that several months ago, Misuto deemed an Adventure Mode speedrun under 2.5 hours being "virtually impossible" due to RNG, and I find it somewhat satisfying that there are now two runs on the board under 3 hours and one that is very nearly under 2 hours. That isn't meant to be a criticism, but just a comment on where we used to be and how far we've come.

@MR.Reese If you want to do it, go for it. I'm working on getting back on top of the Adventure Mode category, but if you set a time for a Crafting% run, I think I'd give it a go too. Sounds kinda interesting.



Actually, after a lot more testing on stream after my record was beaten we found how easily hackable these records are. It's trivially easy to seed a run, including adventure mode. We tested even longer to try to find a way to show if a run was seeded or not but we failed. So really any run could be seeded.

For a lot of circumstantial evidence, go look at Vanakyria's YouTube page, where (s)he has posted world record after world record for all categories in one attempt all done on single weekends and is a new account and has never streamed or been part of the DS community. 6 world records his first weekend after making his account. 11 world records in one weekend later on. then 6 more over the next two weekends. A brand new account posting no PBs and SHATTERING 23 world records in a game that is trivially easy to hack? It's suspicious to say the least.

Honestly all runs should be called into question at this point, even mine (though it was done with a live audience sufering with me through bad run after bad run), it could STILL have been seeded and there's no way to prove it wasn't.