Hm, if you're improving spawn you'd eventually need to improve defenses. It's one of those things, if you made x spawn better it would effect x differently so you'd need to change x and x would effect x differently. Inevitably you'd need to pump defenses slightly to manage the difference.

Besides, if you're making spawn tougher why not make defenses tougher too? More often than not you can practically run through defenses without worry(top base for instance).

this is what i believe they are discussing about regenerating cannons to make them better vs spawns especially as the game goes on.



Hmmm Idk how you figure this at all. But if switches do tend to stick with one way or another that would explain things during my testing.

When i did 10000 randomizations of a switch i found it to be a slight variance but more than it should be if it were actuall variance. out of 80000 it was off by about 2.5% at the worst. To me this seems quite erronious as i would expect out of 60000 dice rolls it to be something like 10100 sixes, 9900 ones, ect It shouldn't be that far off.

8/10 seems a bit too high from my dice roll results randomization that i ran. Btw i used an 8sided dice (3 switches). Different numbers were at the top each time i simulated 80000 rolls. I never even saw a biased variance either. where 4 dice sets were generally higher than 4 others. The results were something like 10250, 10130, 10110, 10070, 9990, 9850, 9820, 9780.

What I was thinking for a method of randomization, is that you do a combination of DCs and switches.
A constant loop of DCs is running, each number represents a selection (15 heroes, 0 - 14 DCs)
This would make it really hard to choose the same character in a row, unless you rushed to the beacon and got to it at the exact same trigger loop.

However, to make it more random in case said scenario:
A number of switches (more switches = more random) are constantly randomized every trigger cycle.
Some switches may add a DC, add a few DCs, subtract a DC, depending if on or off.
This way, each second a player waits, the selection can randomize many times over.

What I did for my randomization is have one trigger that constantly randomizes a switch. Then I roll a DC 0-12 and each DC represents one hero when switch is set, and a different hero when switch is clear. I even had it so that if you random an already chosen character, you just repeat the cycle again and the switch and the DC both will change.

See? I've asked this Randomization Factor question before, but most ppl just said I was paranoid..

S'funny, because I also randomized Dark Mage 4 times in a row before... =\ ... I think I absent-mindedly made it so that in my later versions, the switches are randomized always... but I'm not too sure, and it wouldn't be much help...

Rather than theory-crafting, Ima make a map where a switch set = +1 mineral and a switch cleared = +1 gas... I'll report back soon

Okay, I have been tinkering with my Random Switch Tester map.

What I did was that whenever the switch is set, it adds an ore for me, and whenever the switch is cleared, it adds a gas for me.
I've played 20+ games of it, and from what I see, the ore and gas amount are usually within 10 of each other all the way up to the 1000s (denoting no bias), with the exception of Player 1 and Player 2.

With Player 1 and Player 2, there are inconsistencies, with all the random outcomes as you can think possible.

For example, sometimes Gas would shoot way above ores in the beginning, and the distance between gas and ore amount remains the same. Sometimes it's the ores. Sometimes, it does not happen in the beginning, but around the 50's..
One time, the gas overtook the ore in the beginning, but around the 600's, the Ore then overtook the gas... and I usually think of overtaking as keeping a constant distance of at least 20 units apart.

This, as far as I can tell, does not happen with Player 4 or 5... I only had limited testings to the other players... (and p8 was a mandatory cpu :C)

I've posted my map below..
EDIT: Actually, it may be better if sm1 moved this to the UMS Theories and Ideas section? D: .. this may be getting off topic.. kkthnx

You're not going to get conclusive results when you randomize them thousands of times in the same game, obviously Blizzard isn't going to use a faulty long run algorithm. There may be inconsistencies at the start of a game however, because the randomization happens at nearly the same time every game and I believe that's what the randomization is set to run off of.

You need to test the randomization triggers that you actually use at the start of the game:

Add 1 mineral for the first switch, 10 for the second switch, 100 for the third, etc. Then you can test that with computer players (show leaderboard) and tally the totals to see if you get a lot of the same result for the same players often. That's a lot more work, though.

Post has been edited 2 time(s), last time on Sep 30 2009, 12:20 pm by FaZ-.

faz that method is impossible to get a large enough sample by doing this with one person. Essentially he has to create 100 games in a row and check the switches each time. 100 also is not near big enough sample either.

to complete this one would need a team of people and they would each have to create 100 games. the larger the group the better. 100 people would provide an ample sample but 10 would give a fair anyalsis for the most part.


either way its quite rediculous when you figure this test may not turn up anything besides people being paranoid. If it does turn up something than it was worth it but imo, it would be better to devise a different randomization method for those who feel this randomization is biased. This would save both time and remove doubt of randomization by creating a method that we know will work.

Also, we don't know that any method will work. I'm just saying that if Unholy actually took the time to do his test the results would be completely worthless because he wouldn't be testing the same problem. I don't really think it's a big deal, I've never noticed any anomalies.

I have. Whenever I go random and really want summoner, I get archer.

Last night I played a quick game, and received MUTANT upon my selection of random. Therefore I guess we can write this off as merely a fluke that I was cursed with lame Dark mage and Medic for so many games in a row.

It really depends on your slot, I like to play as green (map camoflague, lol) and I notice it tends to give summoner quite a bit, whereas, when I play as the top team, I never rolled a summoner.

If you think green is camo, you should've seen it as brown before I changed it.

Nice, moose. Even though you totally bailed on me yesterday, I will give you a cool suggestion about player colors. What I often do for my team maps is set up player colors so that one team is "Warm" colors, and the other is "Cool" colors, so that way teams can more easily identify friend and foe. For example, Red+White+Yellow+Orange are good colors for a team. Purple+Blue+Green+Teal are good colors for the other team.

Eh, the warm colours are easier to see on the minimap. Not really much point.

Or just shift+tab in game and you're done without even editing maps.

I've always had a problem with colors on team maps. They always go some wierd pattern or one that doesn't mesh well. Dstrike and SCW have been the best i've seen using either blues and yellows or blues and greens. I really hate black color even though it looks so cool you can never see it coming at you.

anyway for switch patterns and map colors i, like unholy, believe were going a bit off topic. Honestly you guys hate it when we argue over things but i kinda hate it when nothing important is being talked about. Idk though importance is a matter of opinion but I, myself don't think what char you randomize and what color your team is matters a hell of alot when looking at a game.

I like to separate teams by warm vs cold too, but using the neutrals help keep more similar color, like reds vs blues.

Which works great for the enemy Mutant running in with the Zergling wave. Or when he's on the same team as a summoner.

Also, make the neutral more distinct if you do change colors. Tan looks pretty much like yellow on the minimap. I think brown would be more distinct, even with the dirt terrain. Black would probably work well too, if you did that.

Indeed, player color is an often overlooked factor of a map that is pretty important in a lot of ways.

As for neutral colors, I think the default pale blue is always a good choice, especially on dirt terrain.