The new RNG.

Unknown

Hello, It seems that the tiles are dropping using a new algorithm.
As I understand colors come into "waves", say like 4-5 colors are preferentially chosen on a rolling window.
The end result is, we get a (comparatively) very unstable board.
The main concern is that special tiles, especially countdown tiles, are severly nerfed (as they are very quickly dispatch).
The characters who are heavily buffed are those with direct damage, stun and especially board shaking abilities.
I don't know if someone has already made a rundown on the change in characters balance at this point.
Could you share your analysis on the subject ?

Unknown

http://www.d3pforums.com/viewtopic.php? ... 99#p220699
http://www.d3pforums.com/viewtopic.php? ... 50#p243850

The term 'Waves' is only used during battles against > 3 opponents where multiple waves come at you one after each other with no break and no chance to switch heroes or heal with health packs. It doesn't refer to tile generation color.

Unknown

daibar wrote:

http://www.d3pforums.com/viewtopic.php?f=7&t=15895&p=220699#p220699
http://www.d3pforums.com/viewtopic.php? ... 50#p243850

The term 'Waves' is only used during battles against > 3 opponents where multiple waves come at you one after each other with no break and no chance to switch heroes or heal with health packs. It doesn't refer to tile generation color.

I think he was just using it to describe the fluctuations of tile densities for lack of a better term. However, none of us know for certain how tiles are generated. I've seen talk that there's essentially an invisible "board" above the visible 64, and any drops will come directly from this "board." Apparently this was the method in some older Puzzle Quest games, so it's not unreasonable to presume it may be the same.

In any case, I'm not sure if I see exactly the OP's point. If only 4-5 colors are generated each turn, the top half of the board would be less stable, but it should have little impact on the remainder, unless the effect is prolonged. In fact, the bottom half should become slightly *more* stable, since there's a lower concentration of colors that can be matched.

I've noticed that in some cases, when the board is starved for a certain color, it *really* doesn't want to add new tiles of that color (mostly while trying to heal Daken on prologue missions). However, I'm not sure if this is really severe enough to make the board "unstable".

Probabilistically speaking, you should expect the probabilities of none of a certain color over a certain number of new tiles generated to follow a geometric distribution.


(Courtesy of WolframAlpha)

As such, even if the lack of a certain color is present, it may simply be due to the fact that it's a rare occurrence you happened to experience. Furthermore, I haven't noticed any change in tile generation recently (though that's not to say that there hasn't been one). Could you share your observations on this "new algorithm" in your experience?

Vhailorx

Hard to differentiate between this seemingly good faith claim of a new algorithm for tile drops and the weekly "ai is cheating thread" (at least in terms of evidence if not in tone). All the more so because i don't think we players actually know how the tiles are randomized for drops (unless someone like turul has been poking around in the game code on pc).

My general take is that the devs say it's random, and anecdotes aside, we don't have much actual data based upon which we could disagree. If someone wants to collect and organize enough data to really discuss this topic, then great. But until then, it's basically all speculation.

Unknown

As it is random, it is difficult to have confirmation unless you have a lot of samples.

Actually, it is a trend that I noticed a few days ago. I felt there was much more opportunities for 5 matches than usual and that special tiles had more difficulty to stuck.
At first I thought it was bias, or just a bad day.
So I decided to do another try the next day with a fresh mind. And the same thing happened.
I told myself it was still bias, so instead of giving to frustration, I decided to play a team with strong direct damage.
The idea was that I would have equally tough time, which would alleviate the bias.
It turned out that, this time I still got many cascades, but in my favor.

That was strange, so I decided it might be more than bias. But it had to be tested.

There are few easy ways for the developper to force more opportunities for cascades. One is to have a larger board (I'm not to sure about this), the other is to have fewer colors on the board. (The extreme being a mono color board, of course).
I went for the hypothesis they could have gone for fewer colors. The wave thing is how it could be done while still having 7 colors in play, so that is the hypothesis I decided to test.
The last two days I played with that in mind, building my team around that idea and monitoring color predominance on the board.
It went surprisingly well.
It still only two days, and a larger data sample is still needed to form a definitive conclusion.

If you're interested in making more tests, you might want to use a team with a strong board shaker.
A fast board turnover is key to noticing color waves in a way accessible by the brain.

As to why the dev. would want to improve opportunities for cascades and especially match 5, I might point to Jean Grey.

scottee

The devs said tile dropping is random, and I believe them. Afterall, the dev team is really good at using the RNG.

But seriously, it's probably confirmation bias + variance.

Unknown

FrenchTomato wrote:

The main concern is that special tiles, especially countdown tiles, are severly nerfed (as they are very quickly dispatch).

HA! i just had "a brief intermission" node with only Blondie left. she passed matching TWO series of xDP purple. that was a whole lot of matching and avoiding (and no accidental cascading). sad.

Unknown

The devs have said tile generation is handled by rolling a theoretical d7. 6 colors, and TU tile.

Only time this is different is in prologue before team ups are introduced, in that situation, the TU results in a red being placed instead. This is so new players can go "PEW PEW" with Iron Man more often.

PuceMoose

FrenchTomato wrote:

As to why the dev. would want to improve opportunities for cascades and especially match 5, I might point to Jean Grey.

I think that the bit I italicized is the important part - with match-5's causing noticeable effects, you're more likely to remember/remark on such occurrences when fighting with/against Jean Grey. It's similar to how I can't believe how many match-4's I see (on both sides) during a Loki LR/PvP. I don't think the number of 4/5 matches is significantly different, but it seems much more noticeable due to something happening when a 4/5 match occurs - especially if there's one such effector on each team.

Unknown

I agree that it is probably bias + variance. But:

* Don't you feel more confident in your capacity to pull a "Black Bullseye" off, recently ?

I know a few weeks ago I would have dismissed that as not credible. I'm not so sure now.