Cascades without end

MissoesRicRose

I do not know which category should put this topic. I even thought about putting in bugs.

I do not know if it's extreme bad luck or it really is bug. But now opponents make "Cascades infinities."

In 2 matches the game made an infinite move that ended the match. Maybe the difficulty of the game is now super-exaggerated.

Quebbster

When the opponent get it, it sucks.

When you get it, it is awesome.

For some reason a lot of people only remember when the former happens...

Malcrof

Check this out from way back in the day. It definitely goes both ways:

https://forums.d3go.com/discussion/comment/243366#Comment_243366

Borstock

Why do you think the AI is only allowed to fire a power once per turn, even if they have enough AP to fire it multiple times? 

Endless cascades stink, but they do happen both ways. The difference is that you can do more with it. You have the advantage. 

qandols

Stuck boards with unreachable enemy tiles are equally annoying.

Rosraf

Look, it is clear the color drops are not random as of the moment they drop, and the AI calculates them before they appear on a screen. We know this because visual effects from a cascade (AP count, power ups, invisibility, etc) appear before cascade fully appears or completes.  Sometimes the effects of a chain appear way before the chain completes (like 4 or 5 subsequent matches in advance).

So the only question is if information for tiles that have not appeared yet factor into AI’s matching.  The collective experience of many players for years suggests it does. 

I lept track rack of bonus AP (additional AP gained solely by matches made from new falling tiles) in Clash of the Titans matches this morning. AI beast me in bonus AP by 68%!  -And I do try to be mindful of bonus ap drop potential. 

For sure, there are built in advantages for players. But bonus AP seems to disproportionately favor the AI. 

Rosraf

For sure, there are built in advantages for players. But bonus AP seems to disproportionately favor the AI. 

That could just be that the AI is infinitely better at calculating the odds for all drop match potential, and hence gets better results.

bbigler

Look people, it's a machine, it doesn't "think", it runs a program, which is written by developers who make algorithms to randomize color drops and determine color matches and cascades.  The developers were not thinking in terms of winning, but just writing correct algorithms.  The AI knows the results of the drops before you see them because of the delay in graphics processing. 

In the beginning of this game (if my memory is correct), the AI was allowed to fire powers again and again if they had the AP.  So, cascades back then were more deadly.  Things are much better now.  Enemy cascades are purposely limited (see previous thread on the subject) AND enemies only fire powers once per turn AND in a certain order.  For example, if the enemy 4*Carol has 5 yellow and 8 Black, they will place the Black CD tile first and then wait until the next turn to instantly resolve it, because of its order of operations.  This limitation has saved me many times from Carol nuking someone. 

Jaedenkaal

Rosraf said:

Look, it is clear the color drops are not random as of the moment they drop, and the AI calculates them before they appear on a screen. We know this because visual effects from a cascade (AP count, power ups, invisibility, etc) appear before cascade fully appears or completes.  Sometimes the effects of a chain appear way before the chain completes (like 4 or 5 subsequent matches in advance).

Sorry, this is misinformation. Obviously your phone has to randomly calculate which tiles will drop before it graphically shows you what they are.

Rosraf said:

So the only question is if information for tiles that have not appeared yet factor into AI’s matching.  The collective experience of many players for years suggests it does. 

The collective confirmation bias of some players, I'm assuming you mean. Humans are just bad at recognizing random patterns. It's no one's fault.

Rosraf said:

I lept track rack of bonus AP (additional AP gained solely by matches made from new falling tiles) in Clash of the Titans matches this morning. AI beast me in bonus AP by 68%!  -And I do try to be mindful of bonus ap drop potential. 

Even if I assumed that you did this correctly, that's not a very large sample size. And in any case, the state of the board before new tiles drop is not random. ONLY the new tiles that drop are random. 

cpeyton3535

This enemy advantage in drops was a 'perception' that was so prevailing that steps were taken to address the issue.  At least for PvE.  Meaning it is now semi-random.

https://forums.d3go.com/discussion/73325/enemy-cascade-limiting-6-12-18

Didn't seem to work!

Brigby

cpeyton3535 said:

This enemy advantage in drops was a 'perception' that was so prevailing that steps were taken to address the issue.  At least for PvE.  Meaning it is now semi-random.

https://forums.d3go.com/discussion/73325/enemy-cascade-limiting-6-12-18

Didn't seem to work!

Keep in mind though that those changes don't apply to Versus Tournaments (if OP was playing in one, that is), as that would've impacted the effectiveness of defensive teams.

TPF Alexis

They do happen both ways. I've been getting some absolutely ridiculous cascades in Fresh Cut. Like "firing multiple powers before the AI has had a turn" ridiculous.

And yeah, the human brain isn't very good at dealing with probabilities, especially when they start getting in the numbers required for multiple sequential evolutions of an 8x8 grid. For one of the possible-to-calculate examples, if my math was right, there's a 4.375% chance of any given three tiles dropping in all being the same. I've seen 12 moves tossed around as an average length of a match. If that's correct, a match should feature a drop like that roughly as often as not. Now think about how many moves you make over the course of any given event. 70+matches per day for a regular PvE means hundreds of individual moves. That's enough to make some events with truly minuscule odds of happening on any given move quite likely to occur at some point during that event. When you also take into account the number of people playing, literally one-in-a-million events are likely happening to multiple players every day.

Rosraf

cpeyton3535 said:

This enemy advantage in drops was a 'perception' that was so prevailing that steps were taken to address the issue.  At least for PvE.  Meaning it is now semi-random.

https://forums.d3go.com/discussion/73325/enemy-cascade-limiting-6-12-18

Didn't seem to work!

I wasn’t aware that change was made, but I’m glad they did it. Thanks for the link. 

Yepyep

Rosraf said:

Look, it is clear the color drops are not random as of the moment they drop, and the AI calculates them before they appear on a screen. 
...
For sure, there are built in advantages for players. But bonus AP seems to disproportionately favor the AI. 

More telling than anything, is that there IS an algorithm ordering drops. Period.

1. How many times have you experienced a cascade starting a game? The starting frame is engineered. Period.

2. We start the game with the initiative, always; i.e., humans always make the first move. Good right? Well, not always -- we never get a pop reaction to the first mover's action. Think: Je4n. Always, Je4n -- how often otherwise do you get a 5-match to start the game, hmm? It is definitely often that happens with Je4n.

Just these two truths are proof (to me) that there is a process to the drop. It is not random. If the drop is not random -- then what is?

HoundofShadow

Here is something that might be interesting:

https://www.demiurgestudios.com/engineering/a-worthy-opponent/

After they made changes to limit cascade in PvE, I say that it's not as random as before.

qandols

Something hilarious happened on the Jean Gray node for round 6 of Apocalypse. Enemy team was Hood, Loki and Mystique. Thought I was in for a lot of AP stealing and grief. Enemy teams first match was a match-5 that I could set up for them. The ensuing cascades resulted in two more match-5 killing Loki and Hood. Fun times indeed.

Yepyep

HoundofShadow said:

Here is something that might be interesting:

https://www.demiurgestudios.com/engineering/a-worthy-opponent/

After they made changes to limit cascade in PvE, I say that it's not as random as before.

Random is conceptually binary: there is random or not random. <random = not random.

I am neither a mathematician nor a statistician, so perhaps I am wrong. But I don't think that I am wrong about the above. Are any of you women and men out there able to confirm or disconfirm me on this, please? (Be gentle, lol)

sambrookjm

Yepyep said:

HoundofShadow said:

Here is something that might be interesting:

https://www.demiurgestudios.com/engineering/a-worthy-opponent/

After they made changes to limit cascade in PvE, I say that it's not as random as before.

Random is conceptually binary: there is random or not random. <random = not random.

I am neither a mathematician nor a statistician, so perhaps I am wrong. But I don't think that I am wrong about the above. Are any of you women and men out there able to confirm or disconfirm me on this, please? (Be gentle, lol)

https://xkcd.com/221/  Your random number is 4, as generated by a fair die roll. 

But seriously, if you want to get super technical, for non quantum-computers, there's only "non-random" as the random number generator is another algorithm which uses some sort of number from the computer memory and does some calculations on it.  These calculations depend on the algorithm used to generate the random number.  True randomness is really hard to do on a non-quantum computer.  An article about how Cloudflare creates a (supposedly) truly random number is here:

https://www.popularmechanics.com/technology/security/news/a28921/lava-lamp-security-cloudflare/ 

It links to a YouTube video that explains about how real-world Lava Lamps are used to generate random numbers.  It's far more important for them than it is for Demiurge to have real random numbers.  For the purposes of this game, quasi-random is usually good enough...until the AI gets 70 AP in one turn and proceeds to destroy you.

AXP_isme

There’s a story, probably apocryphal, that apple rewrote the shuffle algorithm in the early days of the iPod because it would sometimes play the same song more than once in a row. People don’t really cope very well with true randomness. We’ve evolved a very strong pattern matching capability that tends to make us see patterns even where none exist. 

Phumade

Yepyep said:

Rosraf said:

Look, it is clear the color drops are not random as of the moment they drop, and the AI calculates them before they appear on a screen. 
...
For sure, there are built in advantages for players. But bonus AP seems to disproportionately favor the AI. 

More telling than anything, is that there IS an algorithm ordering drops. Period.



Just these two truths are proof (to me) that there is a process to the drop. It is not random. If the drop is not random -- then what is?

I’m not sure I understand your point?

is the drop random?  No they told you the gems are weighted to drop more often than TUs. By definition that is not random

are tiles randomly placed on the board?  Yes randomly enough that you would need a large number simulator to prove that any one position on the board has a higher affinity for any one tile over another position.

Your question is like asking if every seed in MIcrosoft freecell is beatable.  The answer is no there are unwinnable seed games.  Does that make the game inherently unfair?

AXP_isme

Phumade said:

Yepyep said:

Rosraf said:

Look, it is clear the color drops are not random as of the moment they drop, and the AI calculates them before they appear on a screen. 
...
For sure, there are built in advantages for players. But bonus AP seems to disproportionately favor the AI. 

More telling than anything, is that there IS an algorithm ordering drops. Period.



Just these two truths are proof (to me) that there is a process to the drop. It is not random. If the drop is not random -- then what is?

I’m not sure I understand your point?

is the drop random?  No they told you the gems are weighted to drop more often than TUs. By definition that is not random

are tiles randomly placed on the board?  Yes randomly enough that you would need a large number simulator to prove that any one position on the board has a higher affinity for any one tile over another position.

Your question is like asking if every seed in MIcrosoft freecell is beatable.  The answer is no there are unwinnable seed games.  Does that make the game inherently unfair?

I think you may be confusing randomness with fairness. They’re not the same thing. 

When either team makes a match the game calculates, in a pseudo-random fashion, what gems will fall from the top to replace the missing ones. That happens in a split-second but it takes time for the graphics to catch up, for your eyes to process it, for the messages to make their way to your brain and for you to react. On the AI turn (i’m not sure AI is appropriate since it seems to have little by way of intelligence, but I digress) it’s not waiting for the graphics to catch up before it works out what to do next. There’s almost no lag to work with so all the actions can, and are, queued up and executed sequentially with no gaps. That can give the appearance of cheating - it looks like the AI knows what’s going to happen next. In fact, that’s not the case. All that has happened is that the screen, your senses and your brain are much slower to process these things than the code when it’s following simple steps. 

I get it. It can be really frustrating when you’re on the wrong end of a massive cascade. I’ve had the enemy turn last so long that my phone screen has turned off. When it happens it makes me furious. On top of that you always remember the bad examples and not the times you won on the first move. That’s perfectly normal. You just need to take a deep breath and carry on or, better still, go do something else instead.