Messages

1

Feedback and Suggestions / Re: IMPORTANT - Matchmaking drives off new players

« on: February 28, 2017, 12:26:58 pm »

Hi everyone,
Thanks a lot for the thread. 

So with this update, we made another change.  Right now if you queue for a couple of minutes, you will see a button to browse match list.  And it should be fairly obvious, the display is.  If this is too long, we can try to reduce it to say 1 min. 
Another thing we can do is say, if the concurrent number falls below a level, say 80? Then we default show the match list.  This way, when there are more people in the game, matchmaker can still help put people into matches, and when the population is lower, we can have people use the match list.  How do you guys think? 

@Daft, it's probably me who answered the question if the answer was a long and rambling one.  I definitely didn't intend to make excuses or say that everything is fine.  If I remember correctly, I was trying to point out the good and bad of everything in a more holistic way.  Cuz we had just match list before as you guys probably know, and we drove off a lot of new players.  With just match list, and with lower pop as we started in the beginning after our initial launch player traffic, it was also a big struggle, as lobbies were really fluid, with people hopping in and out, and people yelling at each other for not wanting someone in a lobby.  The wait times were exceptionally long because lobbies were hard to fill up and people would lobby hop and join and leave.  That's not mentioning any balance, and we had even more of a balance issue than we have now, even at lower populations.  So I was trying to say that it's not easy, as we've been there before. 

I wasn't trying to make excuses or just do the bare minimum.  I think if we really take a step back, I feel like I can honestly say that we're not the just do the bare minimum type.  We've poured everything we got over the last 5 years making and supporting the game.  And perhaps we're not good enough, but it's not for a lack of trying.  Anyways, not making an excuse, and thanks a ton for the comment. 

So with that said, please let me know if you guys think the solution I mentioned would be good, and we can make it happen. 

Thanks a lot!  Howard

2

News and Announcements / Servers are now back, and Alliance is back up through this weekend!

« on: January 05, 2017, 06:14:15 am »

Hi everyone! Servers are now back up!  And Alliance is now up again through this weekend in honor of Steam's Daily Deal featuring and the extension of the Alliance Loyalty Reward.

http://store.steampowered.com/app/209080
http://store.steampowered.com/app/252590/

Sorry for the inconvenience! 

Thanks so much, enjoy, and Happy New Years!

Howard

3

News and Announcements / Short Server Maintenance Now for an Hour

« on: January 05, 2017, 04:58:29 am »

Hi everyone! Servers will be down shortly and be back up in about an hour. In honor of Steam's Daily Deal featuring and the extension of the Alliance Loyalty Reward, we're taking Alliance back up for people to play in a bit, and it'll last through the weekend.

http://store.steampowered.com/app/209080
http://store.steampowered.com/app/252590/

Thanks so much, enjoy, and Happy New Years!

Howard

4

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 04:46:37 pm »

Ok understood.  In summary, the areas that we need to pay close attention to are:  spire, squid, and loch. 

5

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 04:33:58 pm »

@richard, oh good question, sorry this was not make clear.  Yeah in a way.  With usage and win rate, it is to see the distribution of usage rates of different ships and builds and see how they are relative to the mean.  It's pretty similar to how say Valve balances CS:GO.  I mean, how they balance it in actuality isn't my point, it's more like how they look at stats and use them to deduce starting point and measure goals etc. 

6

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 01:56:07 pm »

Guys, you can absolutely criticize us for not communicating better, for not implementing all you guys' ideas, and for failing at the actual changes in some way.  That's totally fine.  The only thing I want to reiterate is that, we did not somehow ignore all the feedback and all the ideas.  That did not happen.  I just wanted to reassure that.  I'm not sure how much consolation that is, but I feel like we take feedback really seriously.  I think if you guys go back to our build history, you can find ideas of all kinds being implemented.  I have to check, but we might be over 1K features or changes implemented based on player feedback (don't quote me on it).  This might not sound huge in DOTA's scheme of things, but I for one am really proud of it.  So if you can help it, please don't feel like you are being ignored. 

Even with the changes with squid and spire that you don't agree, we're not drawing a line in the sand or setting anything in stone.  I do hear your points, and we do consider them. 

Thanks a lot everyone!

7

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 01:46:03 pm »

@nanoduckling I have no doubt you forget more about physics than I'll ever know.  All I am saying is that, from a game development point of view, it is not broken.  When you say broken, that to a game dev means there is a bug.  So I was trying to tell you that there was no bug. 

With pyra, I guess it's a difference in thinking on how to approach this.  I feel like a month by month type of shorter iterative adjustment is something I want to avoid.  I don't think that's enough time, but a year is probably too long.  I think the better approach would likely be somewhere in the middle.  And this is a point that some others' have raised as well.  Specifically with the pyra, we did nerf the hull of the pyra slightly.  With the role of the pyra, it was perceived as overpowered the patch before, and the data reflected that.  So we adjusted the pyra.  I think the valid critique of the pyra was the lacking of a more definitive role.  This was something we tried to address in this build.  If we had adjusted the pyra back along the same axis as before, then that oscillation in perception and usage would likely hold true, so with this change, it's more of an attempt to define more of a role.  It seems like it's decently perceived so far. 

Actually, historically the problem has indeed been magnitude, which we've tried to dial down.  Meaning, in the past, we have made changes of greater magnitude, and the over-reaction had been greater.  In this update, I do believe that the actual changes were smaller in magnitude. 

For statistical models, it's really not terribly complicated.  We just want to make sure we don't make decisions anecdotally, and observe emergent patterns, and make sure that when we make a change, we have a way to measure if that change is effective or not.  We do more complex regression analysis for the match system (and I'm sure I'll open a bigger can of worm for mentioning it ). 

8

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 01:07:32 pm »

@zanc, definitely noted.  With balance, in the past, we were more prone to make shorter, quicker, more knee-jerked decisions that were more earth shattering.  Whether they be artemis, heavy flak, pyra, banshee, etc etc, the changes were more drastic, and the cycle of change was shorter.  That was generally perceived very negatively we felt, so we tried to slow things down a bit.  Use data collection to run for longer, try to find trends or patterns.  Ex. with pyra for example, when we looked at the data of both matches overall and vet matches, we saw that, after a perceived nerf, we saw noticeable decline in usage initially.  That's understandable because there was a perceived and real nerf.  So initially usage trended down, and then over time it trended back up, as more people played with it, and eventually reached more of an equilibrium.  If we had not given more time, we might have jumped on more changes right away, and that wasn't the best we felt either.  In this case, have we waited too long?  Perhaps.  It's a bit tough to find that right balance, and we'll try harder. 

9

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 01:01:21 pm »

@dementio, I don't think we've ignored competitive at all to be honest, as a lot of feedback, including yours have been taken into account. 

Right now, you have specific issue with squid and spire, and I get that.  In this case, I do agree that we were motivated by stats in vet matches as well as matches that were predominantly played by vets.  We wanted to get a larger sample size over a longer period.  If this has misled us potentially, we are aware, and these 2 items are things we will be monitoring most closely over the coming days. 

As for trying different ideas.  Granted, we have not tried every single variable that's potentially in existence, surely.  As you guys noted in this thread, we have not been stuck on 1 set of ideas either.  With changes, we had to hold some things constant, and I don't think we ignored the fact that our first ideas could be flawed. 

Well in any case, it's not my place to change your opinion or argue.  I do agree that we have rooms for improvement, and we will look specifically at the spire and the squid. 

Thanks for listening, Howard

10

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 12:49:32 pm »

Just one note on gravity change, the faster fall would have been closer to actual real physics, so it was not broken.  We wanted to see if trying that would allow people to get out of locking and arc faster, but obviously that didn't work. 

With trying the as big as we can in terms of changes as a starting point, we did try to be more careful.  This was because in the past, larger changes were generally perceived very negatively, but it's a fair point. 

With the pyra, if we broke it, it was because it was generally considered significantly overpowered the patch before last, and the data reflected that.  We tried not to make knee-jerk changes and wanted to let data collection last for longer.  Too long?  Perhaps, but I don't think people accept knee-jerk changes now. 

I don't believe there was a range reduction bug with the hwacha.   

11

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 12:20:22 pm »

@Omni, I felt like in the beginning of the process we did, but once again, it just points out the communication gap.  With the changes, we also had to be careful about making drastic changes.  Based on past experiences, it was not advisable to make wild changes, as they would be arbitrary.  I'm not saying that we're more conservative with ideas, but I am saying that we are more careful now than before.  We did try things that were more drastic, ex. the fall etc, in the first two weeks, but those had to be pulled back or discarded, which was fine. 

12

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 12:15:40 pm »

Hi everyone,
There are some back and forth discussions on the balance changes themselves, but let me focus for a moment on the process and the point about player feedback feeling like a waste of time.  Like I said earlier, data, especially data from higher level play (matches with predominantly vets) over the course of time, gave us some starting points.  Over each week of testing, we tried out some different ideas.  We tried the drop, and people gave us feedback on that, and the mechanic was removed.  We tried 3 different adjustments to the mobula guns, and people gave us feedback on that, with which we tried to make the best adjustments possible after each week of testing.  We tried a hwacha firing mechanic and people gave us feedback on that and we discarded.  We eventually settled on a change that seemed to be more palatable to most people who gave us feedback in later weeks of testing.  We had a couple of attempts at the harpoon, and people's feedback grew more positive.  We tried playing with squid's gun angle of different guns, and people gave us feedback on that. 

So I'm not quite sure where the impression of feedback not being taken into account arrived from.  I do want to note that, we did get a lot of feedback, and a lot of them conflicting, which was interesting and good in a way, because people have varying opinions, biases, preferences, etc.  But it does mean that we have to try to look for patterns and people's rationale and logic.  We can't be expected to just take someone's point of view and run with it wholesale I think for obvious reasons, and not taking one person's feedback should not be conceived of as a slight.  Meaning, it doesn't mean we don't appreciate it.  This is the burden with us receiving feedback, is that we had to consider all of them, and it was a challenge for us. 

Having said that, I think we can do better with the communications with each round of testing, and do more to preface the changes before we took the changes live.  We have limited means so far to do that, because not everyone access the same information at the same place and at the same time.  So one though may be potentially to do more announcements in game etc.  We're still thinking about what to do. 

I honestly think we've tried to do better (and the process should have been better than before).  We organized multiple rounds of testing, did blind as well as testing with notes published every round, properly logged all feedback, and had weekly meeting with Matt, me, Alex, Eric, Josie, and Mikko at least to review everything and debate.  Different people's ideas and reasonings were brought up and discussed. 

We definitely have room for improvement though, both in terms of the process itself and providing more context.  We'll try to do better. 
Thanks for listening, Howard

13

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 04:06:11 am »

Aww really sorry for the hassle Letus.  And yeah, you did touch on some of the underlying considerations behind flak. 

I mean, we could start this type of conversation earlier as some of you guys want.  One reason why we did not start this type of discussion openly is so we don't introduce more bias into this process.  It does make communication tougher, so there's a trade off.

14

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 04:02:34 am »

I'm still traveling and completely exhausted, so apologies once again for the short reply or for any incoherence. 
I do just want to quickly add to what Matt and Jub Jub said earlier.  We do look at how things trend as well as usage and win rates in pub matches overall, as well as in novice matches.  Hwacha for example was an issue area that we wanted to address.  However, we looked a lot at high level play as well.  From a data standpoint, if we were to be responsible in looking at data as a starting point to identify issue areas, we had to look at data across a longer time frame and larger data sample.  Heavy flak was an example of this.  The ships we made adjustments to followed the same arc.  Process wise, we did look towards feedback for ideas, and we came up with some ideas ourselves as well obviously.  What was also valuable was the interim feedback that ended up giving us references or ideas for week to week changes.  If we looked at hwacha, we initially experimented with the way it fired, and that didn't go over well, so we started looking at a different solution, and the current solution seemed better received.  We went through a bit of this with the squid.  While some of you are not happy with the acceleration change still, the gun arc changes did go through a few iterations, and these iterations we did do largely based on feedback. 

So I would say that we do make sone considerations for general pub and lower level play.  But we also do make a lot of considerations for higher level play as well.  Different levels of play have different issues, which add to the complexity when it comes to balance.  The actual picture for us has always been more complex. 

So I do agree with Jub Jub, but I think the actual picture for me is more complex. 

15

Release Notes / Re: 1.4.5 New Balance Changes "When Ambush Comes To Shove"

« on: March 19, 2016, 02:28:02 am »

Omni, yeah I know what you mean, and I was trying to say that in the beginning of the process, we did try to do that.  But I do agree that as we released the actual patch and notes, we could have reiterated the context. This I talked to Matt about today and we will do better next time.