Posts on Current Forum | Archived Posts
|
|
|
||
|
||||
|
||||
|
||||
| Frequently Asked Forum Questions | |
|||
|
||||
|
|
|||
| ||||
| ||||
| ||||
| Search Older Posts on This Forum: Posts on Current Forum | Archived Posts | ||||
| ||||
| ||||
| ||||
| Re: re: DMR Study Update (new spreadsheet) | |
| Posted By: Hoovaloov | Date: 1/30/12 9:22 a.m. |
In Response To: Re: re: DMR Study Update (new spreadsheet) (RC Master) : Well it should do: the method you just described will now just be sampling : from the 625 possibilities. The question you should be asking is: why do : you still have error when you've oversampled? :P No clue. But it's encouraging to me that it was still that close (0.6%) to the true percentages. -----------------------
: Which makes it optimal by 1 frame - lol! Of course we know from the spamming
: in 85%
: Which means that, in theory, spamming 4 body shots and then going for the
So if we exclude full-on spamming as a tactic, and just focus on 4 spam + 1 headshot vs. bursts of 2 + 1 headshot, it appears that 100% bloom encourages the 4 shot spamming technique and 85% bloom encourages the multiple reset technique. Or to put it another way, 85% bloom encourages pacing even within this type of tactic. ----------------------- : What I find even more interesting is how these numbers relate to the
: Mid-range
: Long-range
: So your DMR could use some work Hoov ;) Theres an error of something like 1.5
Yes, after calculating the reset times of the reticule after each spammed shot, I saw how much reaction time error there was in my full paced shots. In some kills, I added up to 15 frames of extra time to the full paced shots. But if we examine the spamming times, there is no reaction time error in those times since I was pulling the trigger much faster than the DMR could actually shoot. (The one or two frame differences come from the random chance of an 11 frame spam shot, something outside of player control). This means I was unfairly skewing the pacing times with my reaction time error. We have removed all other player influences and error from the study (movement, skill, accuracy), but there is still error in reaction time of the pacing. So I propose that because the spamming times do not include player error, we should remove the error in pacing times. In order to accomplish this, we would just use the reset values that I provided to calculate the fastest full reset pacing kill possible in each bloom setting: 100% bloom:
85% bloom:
In other words, replace all of pacing times with the minimum full pacing times according to the number of shots fired. These values have no human reaction time error, and provide us with a better comparison between the actual effects of the bloom settings. If we did this, we would be relying purely on the game's probability to give us a 5 shot kill, or 6 shot, or 7 shot kill, etc. at different bloom settings and ranges. We would have completely removed another variable, leaving us with just the change in bloom setting as the only independent variable. I took your probability matrices, and replaced the paced times with the error-free values. So let's look at the change this makes to Absolute (no lag): Error Data:
Corrected Data:
Not much changed at medium and long, but it seems like the new error-free data is doing way worse at short range then the old error-filled data, right? Interestingly, the 85% bloom paced winning percentage has not changed. It is still at 40% for either the old or new data. But the 100% bloom paced went way up, from 28% (old data) to 52% (corrected data). And now when we compare 52% to 40%, it looks like a massive decrease. Basically, the new data has a large probability of a draw at 100% bloom, but they are gone in 85% bloom, which affected the paced % much more negatively than the spammed %. But we know that Absolute (no lag) is not really the best perspective because it is obvious that every multiplayer game has some lag tolerance. Since we agreed that 3 frames of lag tolerance might be a bit low for real world, I used a lag tolerance of 4 in the updated version of the DMR Matrix. So now let's re-examine the games to 100 scenario: Short Range
Score 89-100 -- spammer wins by 11 85% bloom
Score 100-79 -- pacer wins by 21 Net Change: Pacer gains 11, Spammer loses 21 Medium Range
Score 100-41 -- pacer wins by 59 85% bloom
Score 100-25 -- pacer wins by 75 Net Change: Pacer gains 0, Spammer loses 16 Long Range
Score 100-13 -- pacer wins by 87 85% bloom
Score 100-13 -- pacer wins by 87 Net Change: Pacer gains 0, Spammer loses 0 So now that we've removed my reaction time error from the pacing side of the experiment to match the lack of reaction time error in the spamming side, we can see just how much 85% bloom improves life for the pacer. At short range, not only does the pacer gain 11 kills, the spammer loses 21 kills as well, resulting in an actual switching of the winner of match from the spammer at 100% bloom to the pacer at 85% bloom. At medium range, the pacer wins both matches, and his lead is increased by 16 kills in 85% bloom. At long range, there's no change. These conclusions closely match my original results (big change at short, decent change at medium, not much at long range) And lastly, I updated the Win Margin matrices. What stands out right away with the new data is how at short and medium range, going from 100% to 85% results in the removal of all dark red, and some dark green. In other words, more of these battles are getting closer together and there are less large victory margins. This is to be expected as the time between spam and pacing is reduced. And this actually gives the edge to the pacer again because these Win Margins are so small, that whoever misses a shot first will likely lose the majority of the time in 85% bloom. And since pacing is vastly more accurate and consistent, this rewards pacers. ----------------------- : Opinion time! As for how to re-balance the sandbox, I only hear the complaint of a broken sandbox in regards to automatic weapons, and possibly the pistol. I would much rather see an increase in their damage and accuracy then a nerf of the DMR/NR. TU adds some much needed reliability to the DMR, and the freshness and faster tempo of the game certainly help if flow better. As for the NR, there's really only three options: 1) remove it completely, but that's lame 2) make it a starting weapon so it's fair, but that would basically turn everything into ZB all the time, or 3) make it a new "power weapon" and have only one or two on the map with 2 or 3 minute respawns and shallow clips. Good for a couple kills and then it's done. BTB maps need to take a closer look at player cover in TU gametypes. So wide open maps are pretty much WWI trench lines. Cross the no-man's land and you're done for. I think the Community could step up here and add a few strategic rocks or small buildings to maps like Hemmorhage, Wayont, or other open maps that have become tougher to traverse. Paradiso is beyond fixing, IMO. ;) : On the other hand, if you really wanted to punish spamming, bloom should
If reducing bloom "breaks" Reach, then adding bloom will fix it, right? While this would definitely punish spamming more, it causes more problems than it fixes. Let's imagine what it would be like: Say we significantly increase the bloom of the DMR and make bloom so big that if you ever start to spam, you are almost guaranteed to miss. The problem with this is, since you have to wait a lot longer before the next shot, the game would grind to a halt. The other problem with this is, if spamming is punished this badly, you're basically forcing the player to fire at a specific rate. So then why even allow players to spam? Why not just remove the bloom mechanic altogether and have a fixed rate of fire? Now we're back at Zero Bloom. But obviously ZB is not balanced at all. The only way a ZB DMR would be balanced within the sandbox would be to drastically slow down the ROF. But slower is not the answer. So what is the answer? I think the data shows that the TU does a great job of rewarding player skill, making the game a bit faster, and making it fun to fire the DMR again. The only thing the TU needs is a buff to automatic guns for a bit of balancing, and I think it's the best Reach will get.
| |
|
| Replies: |
The HBO Forum Archive is maintained with WebBBS 4.33. |