Design post on the blog!

[Galefury]

No.

This sentence is just to get to the 25 character limit.

[dislekcia]

This is going to seem hugely off topic, but I actually think it is relevant and would love to hear opinions and actual facts:
Can every dungeon be completed purist every time?(Let's ignore bugs for now, the screwed thread highlights those nicely). Meaning, if a person played perfectly but without additional knowledge could every dungeon achieve a purist win? Similarly can every badge/class/race have a strategy created that will guarantee a win for a given dungeon?

Probably. Mathematically the resource distribution and resource need curves are known per dungeon, so given a brand new dungeon, there is usually at least 1 definitely solvable set of actions a player could perform - given that a generation mishap hasn't rendered an impossible path. Most early dungeons probably have quite a high number of valid solution sets, vicious dungeons might only have an average of 20-30 or so per dungeon generation. You'd have to ask Aequitas for more accurate numbers, his brain works that way

Although that's mostly educated guesswork based on early investigations with the base DD system when I was rewriting it for Unity. I haven't gone very far down the road of building a complete depth-first DD solver, mostly because the game's events stopped being reversible a while back and you need that to be able to step back to previous states if your solver runs into a dead end.

[The Avatar]

I'm not surprised there is one set of winning directions for anything purist. But consider how hard it is to come to that (or even one of the 30. You make a ton of moves with a ton of other things you could do. The chance of hitting the one of 30 solutions acting randomly, or even non-randomly, is tiny. So yes they can be all won purist always. But it will take a lot of tries to even pull that off once in a harder dungeon.

[dislekcia]

I'm not surprised there is one set of winning directions for anything purist. But consider how hard it is to come to that (or even one of the 30. You make a ton of moves with a ton of other things you could do. The chance of hitting the one of 30 solutions acting randomly, or even non-randomly, is tiny. So yes they can be all won purist always. But it will take a lot of tries to even pull that off once in a harder dungeon.

The thing that makes the game playable is that any tree of actions that results in a successful run is going to have quite a few points of overlap with other successful and unsuccessful trees. Areas where actions are the same in both trees or one set of actions happens before another in one tree and the other way around in another (imagine the choice of killing two enemies that have no status effects and won't make you level up, the order you do those actions in is almost irrelevant). That ends up turning that huge mishmash of complexity into something more akin to a strange attractor as you get better at the game and steering towards likely winning trees...

But yeah. Math. Kinda...

[Galefury]

Well, I decided to post my thoughts on this in a longer form after all.

Can every dungeon be completed purist every time?(Let's ignore bugs for now, the screwed thread highlights those nicely).

Isn't the "screwed" thread just a bunch of examples of unsolvable dungeons? You declare unsolvable dungeons as bugs, then ask whether there are unsolvable dungeons when ignoring these bugs. This does not make sense. Make up your mind.

Also, I suspect the only way to *guarantee* a dungeon is solvable is to find a solution, and this is just not practical to do on level generation due to cpu time constraints (finding one solution can be fairly fast, finding out that there isn't one can take a while if you're not blocked in right at the start). And dislekcia just said even the devs currently don't have a solver, so they *cant* guarantee that a dungeon is solvable in this way.

The number of winning sequences of actions in a randomly generated dungeon is random, and I cant see any reason why it could not sometimes randomly be 0. The only way this could be prevented is somehow excluding unsolvable dungeons from generation, which as I explained above, is not done. The probability that there is no winning sequence of actions may be small, but if it is not exactly 0, with enough tries you *will* generate an unsolvable dungeon.

So the answer is no.

Edit: vetoing dungeon instances with early blocks is probably possible and practical (for example by checking whether there are paths to a bunch of low level monsters), and might be done. It's probably also possible and practical for some other causes of unwinnable dungeons, and maybe some are vetoed. I just don't know. If this is the case you would have to be really unlucky to hit an unsolvable dungeon.

[xspeedballx]

It is not a design goal to allow for a person to be locked surrounded by level 3's. So it's a bug. A lot of those were basically that in short or long form.

If the game started you at level 10, with the whole level revealed doing 900 damage a hit and with only one enemy at level 1 with 5 hp and 0 damage. You have a guarantee solve. This sounds ridiculous. But Guaranteeing a dungeon is solvable does not require a solver.. it requires balance and tuning and math. I doubt they apply sigma's but if the game was at 4-5 sigma for "solvable vs unsolvable" that would be less than I believe .0001% failure rate. Acceptable by a lot of standards as "working."

[OneMoreNameless]

And we're saying that any resource cost placed on a locker system, be it time, gold, real cash money or sloth pictures, is going to become a "fun tax" for players over time.

"People are going to get bored of our game eventually, so there is no reason to make it any less boring in the meantime."

The current "fun tax" is a lot less time consuming and taxing than any of the previous "fun taxes".

"We have improved our game previously, so there is no reason to continue improving our game now."

[Galefury]

I doubt they apply sigma's but if the game was at 4-5 sigma for "solvable vs unsolvable" that would be less than I believe .0001% failure rate. Acceptable by a lot of standards as "working."

Yes, of course. But that was not the question. The question was "Can every dungeon be completed purist every time?" (emphasis added). Note that "every dungeon" includes unlocking a vicious dungeon with as few other unlocks as possible, and then trying to beat that vicious dungeon with the class least suited to it, purist. I don't know if it's possible to unlock a vicious dungeon without unlocking any gods, but if so, I think it might be possible to (deliberately) create a situation where you would have to get lucky for a specific dungeon to be solvable with a specific class.

This is a very constructed case though, I'm sure by far the most dungeons seen by players are solvable. And if one in a million isn't, who cares. Doesn't change the answer to the question you asked.

Maybe I have been slightly damaged by math lectures.

[TigerKnee]

...What I am seeing is that by offering more control with locker slots and preps in general you reduced the random and replaced it with constants. Now players can SEE clear efficient paths but not always reach them(though they are obtainable) which feels painful. Those players this effect are demanding for a solidifying of the constants. When the reality is the removal of them altogether(locker slots not being restorable going in that direction) would have a similar comforting effect by moving towards what the Devs are describing as playing a run for fun, and getting what you can out of it...

I think I've made quite a bit of "I like Alpha better, but I can't really put a finger on why" posts over time but your post is a pretty good summary.

It feels to me that if you weren't FORCED to play purist to get some badges and other such "artificial" restrictions (to me anyway), playing with preps is better than playing naked in terms of "makes you have a higher chance to win dungeon".

The whole "prep" system is rather problematic to me. I thought a strategic layer would make the game more interesting but as time goes on I think it has the same problem has most games with "skill setups"... the optimal choices are discovered and there's no fun in it anymore (see: League of Legends Runes and Masteries system)

Generally speaking, I'm not the sort of person to do stuff for the sake of saying I did it, and I want the "lots of random/do your best with what you get" gameplay to be the optimal one, and while I can certainly still play DD Beta that way by doing Purist runs... I don't feel very rewarded when doing it.

[Bloggorus]

According to my mathematical education, in a system this complex there is no way of definitively saying that every dungeon is completable purist every time.

Maybe in the easier dungeons you are approaching a probability of 1 enough to say that it's a definite yes. But I'd also say that in vicious dungeons the chance of success approaches 0 to the point where numbers break down, ie. the variables involved make the outcome uncomputable