Discussion in 'Dungeons of Dredmor General' started by Nicholas, Feb 27, 2012.
Okay, okay, I volunteer to be the donated bride. I'm not sure what good it'll do, though...
Wow. My choice of wording was poor to say the least.
Either I am getting worse at typing, or my keyboard is selectively changing out and omitting certain characters more regularly. I spend more time proofreading than typing. And that one still slipped by.
Obviously I meant "Bribe" instead of bride. (And I edited in the correct word too, but by all means, leave your quote up there. It looks good.)
Ten bad Krongs in a row, now that's impressive! But was it all on the same item w/o save scumming? That's where it's fun, seeing if you can make the net result of the item negative
As to the traps, mine seems to average out over time. I've had a decent number of 30% traps disarm on the first try and large number of 45% traps take 10-15 tries. 97% of the time though (and that's a real percent), I have the affinity to just pick it up. I carry a "trap affinity' gear set at all times
Odds often work counter-intuitively. A 45% trap actually has more than a 1% chance of taking 8 or more consecutive tries to disarm it, even if the RNG is working perfectly.
Disarming odds on a 45% trap:
Disarm on first attempt: 45%
Disarm in 2 attempts: 25%
Disarm in 3 attempts: 14%
Disarm in 4 attempts: 7%
Disarm in 5 attempts: 4%
Disarm in 6 attempts: 2%
Disarm in 7 attempts: 1%
Disarm in 8 attempts: 1%
Disarm in 9 or more attempts: ~1%
The rounding gets a little dicey near the end of that list, but the upshot of it all is that 1 trap in 20 _should_ take such extended efforts to disarm that it just feels like something's gone terribly wrong with the RNG. Even if the RNG is working perfectly. You will encounter around 1,000 traps before you find Dredmor (75 per floor on GR), so every play-through should have a couple of "is this working right?" moments.
Not that I'm saying that the anecdotal evidence doesn't seem to support the RNG being plagued with streakiness. I've found some real weirdness with item-spawn triggers in rooms, and recently reported it to dbaumgart. I just haven't conducted the sort of extensive testing to rule out perfectly normal randomness. Unless someone's willing to keep a log of thousands of potential activations of some trigger with a known percentage, we can't really say for certain that the game is truly misbehaving. Until then, gambler's fallacy and confirmation bias could be misleading us. Common sense is not to be trusted in the land of probability.
That's the fun (and curse) of randomness. Getting the same thing a million times in a row is a perfectly acceptable random result. Combine that with the human tendency to remember the horrible runs on a single trap and not the runs of 1-shot disarmed traps, and it will pretty much always seem way off
I played 1.0.10RC9 (Steam) and there are still a few small UI problems, I haven't read the whole thread to see if they were all mentioned before but I took screenshots to try to illustrate. First off thanks for the game etc., I wouldn't notice these things if I didn't have 150 hours into it...
First and easy enough is the crafting window and character window overlap incorrectly. Larger UI elements or not doesn't seem to matter. When you open the Character window, it appears on top of the Crafting window, but:
The Character window titlebar isn't active and you can't drag the character window around like you can other windows.
The Character window is apparently under the Crafting window but visually appears to be on top of the Crafting window. The Crafting window has to be moved (using its titlebar to drag it) away from the character window completely so you can close the character window.
Next issue is weirder so I made a composite image to illustrate, the issue is that your belt item 6 is hard to fill. You can only put items in it if all the other slots in your belt are full. See this imageand try for yourself if you haven't noticed... slot 6 is just hard to fill.
There are other minor things but those were what bothered me the most. I (kind of) assume it's known that some interface elements are cut off at certain fullscreen resolutions when using "Large UI elements" but I'm sure that's a more involved fix and maybe lower priority; I'm at a kind of weird resolution (laptop/1366x768 right now which is not very common.) I just switch out of "large UI" to see how much money I have which is not that bad. I love the large UI mode overall.
The part I do not understand about the traps is that I often set those 45% traps off over and over again before I manage to disarm them. Usually I have 75 or more sneakiness and should be able to walk right over them most of the time. So where is the advantage of attempting to disarm? When I take the time to disarm, I should have a greatly reduced chance to set them off as opposed to just walking over them.
If disarming a landmine, I would take my time and be very careful. DoD punishes me for the attempt to minimize trap harm.
Use the keybinding "C" to enable/disable the character window, and "I" for the inventory. These are lowercase and work even when the crafting window is up at the time.
The first rule of videogames - don't attempt to apply logic to videogames!
The experience points you get for disarming the trap are the advantage, I think. That and, if it's a trap you can pick up, the ability to reuse it to squish a Diggle into chunks of ground beast.
Mmmm, ground beast.
You can pick them up and sell them/skol them/stash a lot of them near a monster zoo and observe the random insanity.
It is clear that the player characters in DoD have ADHD. Thus, they attempt to disarm a mine in a second, which usually yields not-so-great results.
I wonder if the Gambler's Fallacy works for the RNG in Dredmor.
Nah, it's just computer RNGs being weird in general (and often producing weird results out of nowhere, though said results *are* statistically possible), and people remembering weird behaviour easier than normal one.
I still think that the RNG is trying to get even for something I did to bypass its will in the past.
Those 45% chance traps spring far too often to make any sort of statistical probability besides the fact that the RNG hates us.
But you are right, I certainly recall when the statistics do not match the results far easier than when they do. That does not however make it normal for statistical probabilities to be off by a large amount regularly. And they are. Way off.
Better said: Do not apply logic. Period. And never look for logic. It is a puppet of the RNG and thus hates us all.
Personally, I blame the French.
Apologies to anyone here who is French.
But yeah, RNGs are flawed in the fact that they're almost impossible to be truly random, therefor, some issues tend to happen, such as certain things taking a lot more tries than they should.
And Krong's super cursing rolls.
I think we're what, about 15 years away technology wise to make a truly random RNG?
Truly random is a philosophy. Not a reality. We use pseudo random number generators all the time for a million and one different purposes. They are close enough to random for our purposes. Even rolling a literal physical die (Singular of dice? Or am I mistaken?) is in no way random. It is simply difficult to plan and most lack the coordination to roll it to land as they desire.
Current RNGs hash out the time to a certain number of decimal points and pick a series of numbers from those based upon a formula. In reality, the difference between psuedo random and *Random* should be indistinguishable for us.
I studied this years ago while researching encryption algorithms. There absolutely is no way to generate *Random* at all. Period.
As for DoD and random disarm failures and NegaKrongs... I suggest that the next RC has more weight towards the players favor each time they have a negative in a row. Sort of like an invisible buff that makes that 45% become 50% after one failure, and 60% the next time and so on. This should make it seem less like the game is cheating us. To compensate, we could reduce the XP gained when you succeed at the disarm.
For Krong, how about we change it so that Krong curses are also used as an invisible buff to favor for the next anvil used for the same item. That would make it potentially beneficial to keep Kronging the same item even if it gets cursed, rather than throw it away in favor of another better item.
What does everyone think of these ideas?
Oh you can get truly random bits (probably), you just don't normally. There are sites that allow you to connect to them and get a number of random bits based on the decay of radioactive isotopes. Said bits should be truly random, but it's possible they're not. Most of the time people just use that to see an RNG though.
Unix also has a system that fills an "entropy pool" based on user input timings (time between key presses and mouse use and such) so that randomness can be attained without having to seed a generator based on the current time.
Random.org seems to disagree with you on the finer points of that statement. Specifically, here. I could be misunderstanding you or them, though.
That would change the numbers to:
Disarm on first attempt: 45%
Disarm in 2 attempts: 28%
Disarm in 3 attempts: 15%
Disarm in 4 attempts: 7%
Disarm in 5 attempts: 3%
Disarm in 6 attempts: 1%
Disarm in 7 or more attempts: less than 1%
Here's the existing numbers as-is for comparison, so you don't have to hunt back to my previous post.
What you're proposing OmniNegro would be an improvement. However, it wouldn't be a huge improvement, nor guaranteed proof against the occasional freakish trap that takes 10 attempts to disarm. If the "problem" is indeed the RNG acting up as some suggest, then +5% per consecutive disarm attempt wouldn't fix that at all.
It's only going to matter on the repeating "gargoyle" style of trap, anyway, since most traps just go off when you fail a disarm. Currently, a gargoyle trap has a 30% chance of taking 3 or more attempts to disarm. With your proposed change, that would fall to 27%. Again, an improvement, but a pretty small one. It won't change the big picture. Either you've got the skill to pick up the gargoyle trigger automatically, or your best bet is to not try to disarm it at all (unless of course there's some some other unusual or mitigating circumstance, such as a wall or blocker in the way that will protect you from the acid bolt). Let sleeping gargoyles lie.
So I'm not sure it's worth a lot of programming effort. If the devs could do it with just a couple lines of code, great, but it's not something worth their spending more than 15 minutes on.
The above read really was interesting, I've been curious about this for a while, but his point stands; random.org doesn't achieve randomness by programming, but with outside sources (I'm impressed at how many different things you can use for that too).
As far as I know, true randomness solely with computers is impossible at the moment.
And that I'd agree with. But he was also saying:
So I took his other statement to be literal:
I thought he was saying "nothing is random" as opposed to "computers can't do random". Re-reading it, I guess one should parse "generate" to mean "generate on a computer", and thus perhaps I was splitting hairs. Let's assume that to be the case. There's a reason I wrote "I could be misunderstanding you or them, though." in my other post.
My bad. Sorry.
Pointless Tangent alert: Yes, "die" is singular for "dice". And yes, in my experience, most people don't roll dice in ways that are truly random either. For that you need either big hands or a dice cup. I've got one person in one of my tabletop RPG groups who squeezes the die tight and then just drops it to the table. She gets streaks of the same number again and again, and I'd suspect her of cheating if it wasn't almost always unfavorable rolls for her. For a while I tried to coach her to roll more randomly, but she seems to have only two modes: a mini-roll that doesn't tumble or randomize the die at all, and a total spaz roll that flings dice across the room and onto the floor. There's absolutely no middle-ground for her.
Separate names with a comma.