the merits of incremental development

[HardboiledGargoyle]

For your information, the dev stance on cutting down the dungeon's size is that it is a good goal, but should be done slowly and carefully over many versions, lest too much gets cut. But shouldn't you just go for where you believe you'll end up, and re-adjust from there as necessary? It is difficult to imagine that excessive cuts to Lair would make it imperative to bump it back up to 8 levels.

Suppose there is a kitchen and it makes soup. And the soup seems too watery. The recipe uses 100 gallons of water, but it seems like using 50 gallons would make a better soup. You bring up, why is the soup so watery, let's cut down on the water. You're told that it's better to take this process slowly - cut down the water by 1 gallon, try that recipe a few times, and repeat. Because otherwise, you're told, the soup might be too dry suddenly. Over the last 3 years the recipe has been already modified to use 20 gallons less than it used to, so steps have been taken and there is progress. And of course, you ask, why not try to immediately modify the amounts in the recipe to what you imagine would be the ideal composition, instead of inching towards it bit by bit. And here we are.

[KoboldLord]

The impact of many changes are not necessarily obvious without actually trying them. If you suddenly decide to cut the amount of water in the soup you are making by half, it will probably end up far too salty, but you won't be able to identify that problem because your solid ingredients will all be congealed into a mass burned onto the bottom of the pan. Making your soup thicker and less watery actually requires you to juggle multiple variables including the quantities of all your ingredients, your cooking time, your heating element, and the stirring process. Making incremental change allows for better control of the process so you can address complications at a level of severity that is manageable, rather than ruining a pan and not knowing exactly what the confounding factor even is.

[dpeg]

1. Because we have the time. (There's no pressure to deliver a final product by some deadline.)
2. Less Work I: fewer reverts necessary. (Smaller chance for full blunders.)
3. Less Work II: easier to estimate follow-up changes (e.g. if/how to compensate some level cut).

In a nutshell: generally speaking, continuous systems/changes are much easier to understand than discrete/sudden ones. For examples from physics, read up on "non-pertubative" and e.g. "instantons".

And KoboldLord's real soup example says exactly the same things, from the kitchen point of view.

[dpeg]

Forgot another, actually crucial merit of incremental changes:

4. Concensus.

It'd be extremely hard to agree on what the ideal size of the total dungeon should be. Instead of discussing this until eternity, it's much easier to cut a few levels in each release. (Lack of concensus has prevented changes in the past, this is not a hypothetical point.)

[Siegurt]

Indeed, I personally thing 8 levels is a perfect amount of Lair, of course if it was changed to 6 or 10 I would feel like it wasn't right at first, but then I'd adjust to playing it however it was. Just because you believe something is the right way things should be doesn't mean there's consensus on that point. (Even when there is it doesn't always mean you were right

(I also feel like the cut from 4 levels of Orc to 2 was too drastic of a change for me personally, having to navigate through the end vault of orc to get to a cut-off section of orc:1 is annoying when it happens, it was less drastic when it was only part of orc:3 I was missing the money from)

[HardboiledGargoyle]

Ah yeah, I was jolted at first by orc's shortening too. Especially with the contrast that lair is still 8 levels deep while orc has the most interesting inter-level structure of any branch, but is a quarter as large. But then you think about it for a bit, let it sink in, see it in action, and realize, it's alright. With new orc it's simpler and more important (thus actually justified) to find all cut-off sections of orc:1.

[all before]

Just to repeat what gets said whenever this debate comes up: anyone who would put a significantly shortened dcss fork onto an online server would be a hero. It's probably true, as KL says, that massively changing trunk at once isn't a good idea, but HG is right in that devs could learn more about what works and doesn't by seeing how an experimental branch plays out. Also, if some people hate and others love a shorter fork, would it be so bad to have options?