Involve your community in fine tuning balance with Beta-testing Server — Guild Wars 2 Forums

Involve your community in fine tuning balance with Beta-testing Server

I just want to start off by saying thank you ANet for recognizing the need for new strategy and taking initiative to address balance issues. I'm cautiously optimistic about the new systems team and I think new directions can solve many issues with balance changes. However I still felt that there are still significant risk of unable to test new changes, predominantly due to understaffed or insufficient data from internal testing with the new changes. So I think now is the chance to urge you to consider beta-testing servers.

As many players reflected across forums I think the biggest issue ANet faces when balancing classes is establishing new goals for specs/builds and fine tuning the change. Quite often we see two outcomes from new changes: they either doesn't address core issues with a overpower/underpower spec, or classes get double whammy/gifted so much that they reach ridiculous lows/highs. A good example is druid's current undesired state in PvP given old nerfs still lingered after you address some core issue with celestial avatar; Or weaver in PvE raid with its repeated nerfs happening alongside some significant changes to heavy hitting skills. While the intention to "balance so no spec is towering other" is good, quite often players felt you're struggling with fine tuning and achieving the "right balance". This also translate to community culture where meta builds in high-end contents are volatile and players often need to multi-class to be desired in pug, due to mere difference between meta and off-meta.

Given that current balance schedules are usually months (with ANet saying increase frequency from quarterly), I think it can benefit greatly if you involve community in new changes for 1-2 beta testing weekends before each balance patches. This would be similar to what we had back in HoT BWEs, thus giving developers necessary statistical data for fine tuning new buff/nerf. In these beta-testing weekends players can experience and test out newly changed features, what's valuable will be their actions/build decisions rather than online posts. Then base on these statistics from community effort you can fine tune the numbers without changing/rewriting the functions and put extra stress on balance team. I'm sure the amount of information you can generate from 48 hours of community effort is much better than few dozens hard working internal testers, and those overused or underused skills/traits/builds will become immediately obvious to the analytics.

This could occurred in controlled environment for all three gamemodes, where players are limited to beta-server only. For example PvP players are limited to beta-player locked customized arena and PvP lobby; WvW players are limited to EotM; and PvE players are limited to Aerodrome's Special Forces Training instance (with squad features to invite players).

I think two big things to emphasis in these community events is that whatever players see in beta-testing servers are subjected to change, and given the proximity of these testing weekends to actual balance dates it is important to stress that only numbers will be fine tuned. And functionality changes are reserved to hotfixes schedules. In this way players can actually understand ANet balance directions better few weeks prior and give feedback, thus creating a more understanding cycle rather than people feeling left out.

[RED] Crimson Sunspears...your small family guild since 2015.

Comments

  • Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

  • NICENIKESHOE.7128NICENIKESHOE.7128 Member ✭✭✭
    edited August 22, 2018

    @rank eleven monk.9502 said:
    Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

    What I'm suggesting is NOT for them to get feedback from online forum posts etc, but from statistical information on how often a trait/build/class is used and how well they performed during beta testing. There will be some bias ofc, but it'll be smoothed out much better than internal testing given the sample size.

    The metrics are there, as Irenio explained somewhere in this forum unless I stand corrected.

    [RED] Crimson Sunspears...your small family guild since 2015.

  • Ayrilana.1396Ayrilana.1396 Member ✭✭✭✭
    edited August 22, 2018

    @NICENIKESHOE.7128 said:

    @rank eleven monk.9502 said:
    Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

    What I'm suggesting is NOT for them to get feedback from online forum posts etc, but from statistical information on how often a trait/build/class is used and how well they performed during beta testing. There will be some bias ofc, but it'll be smoothed out much better than internal testing given the sample size.

    The metrics are there, as Irenio explained somewhere in this forum unless I stand corrected.

    How would that be any different from what they’re doing now other than increased costs on their end?

  • @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @rank eleven monk.9502 said:
    Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

    What I'm suggesting is NOT for them to get feedback from online forum posts etc, but from statistical information on how often a trait/build/class is used and how well they performed during beta testing. There will be some bias ofc, but it'll be smoothed out much better than internal testing given the sample size.

    The metrics are there, as Irenio explained somewhere in this forum unless I stand corrected.

    How would that be any different from what they’re doing now other than increased costs on their end?

    What we are seeing now is both number and functional changes are put into hotfixes. Majority of time once a new trait/skill is released on balance patch it is determined, if new changes that are found to be overpower/underpower didn't meet the schedule and remained in place for 3 months until the next balance patch. Then we see some backtracking changes such as split/unsplitting or change/unchange the same thing. Beta-testing servers can shorten this time by putting less load on hotfixes, letting players test out new changes before they're finalised. Think of this as breaking down major goal into minor goals for step-wise approach and fine tuning. Same end goal, but more efficient and less time to reach there.

    [RED] Crimson Sunspears...your small family guild since 2015.

  • Ayrilana.1396Ayrilana.1396 Member ✭✭✭✭

    @NICENIKESHOE.7128 said:

    @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @rank eleven monk.9502 said:
    Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

    What I'm suggesting is NOT for them to get feedback from online forum posts etc, but from statistical information on how often a trait/build/class is used and how well they performed during beta testing. There will be some bias ofc, but it'll be smoothed out much better than internal testing given the sample size.

    The metrics are there, as Irenio explained somewhere in this forum unless I stand corrected.

    How would that be any different from what they’re doing now other than increased costs on their end?

    What we are seeing now is both number and functional changes are put into hotfixes. Majority of time once a new trait/skill is released on balance patch it is determined, if new changes that are found to be overpower/underpower didn't meet the schedule and remained in place for 3 months until the next balance patch. Then we see some backtracking changes such as split/unsplitting or change/unchange the same thing. Beta-testing servers can shorten this time by putting less load on hotfixes, letting players test out new changes before they're finalised. Think of this as breaking down major goal into minor goals for step-wise approach and fine tuning. Same end goal, but more efficient and less time to reach there.

    How’s it different from now? Hotfixes would occur in the testing environment just as they do now.

    Testing on beta servers would be no different than testing on live servers except for the additional costs on Anet and that you’re not on the live servers.

  • @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @rank eleven monk.9502 said:
    Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

    What I'm suggesting is NOT for them to get feedback from online forum posts etc, but from statistical information on how often a trait/build/class is used and how well they performed during beta testing. There will be some bias ofc, but it'll be smoothed out much better than internal testing given the sample size.

    The metrics are there, as Irenio explained somewhere in this forum unless I stand corrected.

    How would that be any different from what they’re doing now other than increased costs on their end?

    What we are seeing now is both number and functional changes are put into hotfixes. Majority of time once a new trait/skill is released on balance patch it is determined, if new changes that are found to be overpower/underpower didn't meet the schedule and remained in place for 3 months until the next balance patch. Then we see some backtracking changes such as split/unsplitting or change/unchange the same thing. Beta-testing servers can shorten this time by putting less load on hotfixes, letting players test out new changes before they're finalised. Think of this as breaking down major goal into minor goals for step-wise approach and fine tuning. Same end goal, but more efficient and less time to reach there.

    How’s it different from now? Hotfixes would occur in the testing environment just as they do now.

    Testing on beta servers would be no different than testing on live servers except for the additional costs on Anet and that you’re not on the live servers.

    How often do you see balance hotfixes happening after a patch? Once? Twice? Maybe three times if there's a game breaking bug? Like I said before there isn't enough prior stress test on the new changes (emphasis on new changes), and after a patch hit quite often broken balance didn't get patched in time and we're left hanging for the next 3 months.

    Beta testing gives a buffer for the Swiss cheese risk prevention model. Given the amount of players involved it'll certainly expose obvious and hidden flaws faster than having dozen staff members stress testing it before it is finalised and go live. In essence you can even say less cost are spent on employees using their time to test something players can give immediately. What's good is it is justifiable for beta testing to be volatile, not live testing or live server. Number changes can be made with little time involved after analytics determine the flaw, as oppose to functional changes which is what we usually see in hotfixes.

    Also you seem to vastly overestimate cost of these events. As far as community and development concerns, hotjoin in PvP and EotM are dead game modes, especially with the current reward system. It cost next to nothing to test on them and it only takes minimal work to set it up.

    [RED] Crimson Sunspears...your small family guild since 2015.

  • Ayrilana.1396Ayrilana.1396 Member ✭✭✭✭

    @NICENIKESHOE.7128 said:

    @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @rank eleven monk.9502 said:
    Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

    What I'm suggesting is NOT for them to get feedback from online forum posts etc, but from statistical information on how often a trait/build/class is used and how well they performed during beta testing. There will be some bias ofc, but it'll be smoothed out much better than internal testing given the sample size.

    The metrics are there, as Irenio explained somewhere in this forum unless I stand corrected.

    How would that be any different from what they’re doing now other than increased costs on their end?

    What we are seeing now is both number and functional changes are put into hotfixes. Majority of time once a new trait/skill is released on balance patch it is determined, if new changes that are found to be overpower/underpower didn't meet the schedule and remained in place for 3 months until the next balance patch. Then we see some backtracking changes such as split/unsplitting or change/unchange the same thing. Beta-testing servers can shorten this time by putting less load on hotfixes, letting players test out new changes before they're finalised. Think of this as breaking down major goal into minor goals for step-wise approach and fine tuning. Same end goal, but more efficient and less time to reach there.

    How’s it different from now? Hotfixes would occur in the testing environment just as they do now.

    Testing on beta servers would be no different than testing on live servers except for the additional costs on Anet and that you’re not on the live servers.

    How often do you see balance hotfixes happening after a patch? Once? Twice? Maybe three times if there's a game breaking bug? Like I said before there isn't enough prior stress test on the new changes (emphasis on new changes), and after a patch hit quite often broken balance didn't get patched in time and we're left hanging for the next 3 months.

    And those same balance hotfixes that appear on the live servers would appear on the test servers.

    Beta testing gives a buffer for the Swiss cheese risk prevention model. Given the amount of players involved it'll certainly expose obvious and hidden flaws faster than having dozen staff members stress testing it before it is finalised and go live. In essence you can even say less cost are spent on employees using their time to test something players can give immediately. What's good is it is justifiable for beta testing to be volatile, not live testing or live server. Number changes can be made with little time involved after analytics determine the flaw, as oppose to functional changes which is what we usually see in hotfixes.

    And having it on the live servers would most certainly expose obvious and hidden flaws faster than having a dozen staff stress testing as well as not adding extra costs.

    Also you seem to vastly overestimate cost of these events. As far as community and development concerns, hotjoin in PvP and EotM are dead game modes, especially with the current reward system. It cost next to nothing to test on them and it only takes minimal work to set it up.

    So you’re saying that Anet can make balance changes that only affects hotjoin PvP and EotM?

  • Knox.8962Knox.8962 Member ✭✭✭

    You also seem to completely neglect the idea that, for PvP espcially, if you think queue times and finding groups are bad on Live servers, imagine how long they would be on the test realm if you split the playerbase between live and test.

    Hard to give feedback or test anything seriously when you don't have enough players to play with.

  • NICENIKESHOE.7128NICENIKESHOE.7128 Member ✭✭✭
    edited August 23, 2018

    @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @Ayrilana.1396 said:

    @NICENIKESHOE.7128 said:

    @rank eleven monk.9502 said:
    Community is incredibly biased though (speaking of PvP, but most likely applies to all areas)

    What I'm suggesting is NOT for them to get feedback from online forum posts etc, but from statistical information on how often a trait/build/class is used and how well they performed during beta testing. There will be some bias ofc, but it'll be smoothed out much better than internal testing given the sample size.

    The metrics are there, as Irenio explained somewhere in this forum unless I stand corrected.

    How would that be any different from what they’re doing now other than increased costs on their end?

    What we are seeing now is both number and functional changes are put into hotfixes. Majority of time once a new trait/skill is released on balance patch it is determined, if new changes that are found to be overpower/underpower didn't meet the schedule and remained in place for 3 months until the next balance patch. Then we see some backtracking changes such as split/unsplitting or change/unchange the same thing. Beta-testing servers can shorten this time by putting less load on hotfixes, letting players test out new changes before they're finalised. Think of this as breaking down major goal into minor goals for step-wise approach and fine tuning. Same end goal, but more efficient and less time to reach there.

    How’s it different from now? Hotfixes would occur in the testing environment just as they do now.

    Testing on beta servers would be no different than testing on live servers except for the additional costs on Anet and that you’re not on the live servers.

    How often do you see balance hotfixes happening after a patch? Once? Twice? Maybe three times if there's a game breaking bug? Like I said before there isn't enough prior stress test on the new changes (emphasis on new changes), and after a patch hit quite often broken balance didn't get patched in time and we're left hanging for the next 3 months.

    And those same balance hotfixes that appear on the live servers would appear on the test servers.

    And how would you know that? It has never been done before in GW2 and you're already assuming it'll be the same. With more time and crowd of information they get with beta server they'll have more room to make adjustments.

    Beta testing gives a buffer for the Swiss cheese risk prevention model. Given the amount of players involved it'll certainly expose obvious and hidden flaws faster than having dozen staff members stress testing it before it is finalised and go live. In essence you can even say less cost are spent on employees using their time to test something players can give immediately. What's good is it is justifiable for beta testing to be volatile, not live testing or live server. Number changes can be made with little time involved after analytics determine the flaw, as oppose to functional changes which is what we usually see in hotfixes.

    And having it on the live servers would most certainly expose obvious and hidden flaws faster than having a dozen staff stress testing as well as not adding extra costs.

    Except every time they release a balance patch they have 1-2 weeks to meet the hotfix deadline, beyond that and we're looking at 3 months of overpowered/underpowered trait/skill/build. An extra weekend of stress testing new changes before it goes live will give much more efficient use of staff cost and time than having "internal testing" with these staff.

    Also you seem to vastly overestimate cost of these events. As far as community and development concerns, hotjoin in PvP and EotM are dead game modes, especially with the current reward system. It cost next to nothing to test on them and it only takes minimal work to set it up.

    So you’re saying that Anet can make balance changes that only affects hotjoin PvP and EotM?

    Yes, because GW2 has the mechanics for it and has done it in the past. One is called dishonor that greys out all ranked/unranked PvP, and one is done in Desert Borderland stress test. GW2 actually has the advantage of having the mechanics for them, as oppose to many popular competitve games that needs to set up entirely different server for beta-testing. Yet other competitive games rely on those for new changes, why? When you compare their balance outcome as oppose to our 3 monthly updates and stale/monotonous/disastrous meta complaints and ridiculous representations on overnerfed/overbuffed class you'll understand.

    @Knox.8962 said:
    You also seem to completely neglect the idea that, for PvP espcially, if you think queue times and finding groups are bad on Live servers, imagine how long they would be on the test realm if you split the playerbase between live and test.

    Hard to give feedback or test anything seriously when you don't have enough players to play with.

    And you seem to completely neglect the fact that, I say right from my first post, that this will occur in hotjoins/customized arena or EotM. There are literally no waiting time in those modes as the players participated get no reward. But beta-testing new changes itself is a great incentive for players to get into quickly and check it out, while not directly affecting the balance in live server modes like borderlands/EB and ranked/unranked. While you can say live server population are affected, I would argue that people prefer better balance in long run than having wait extra 1-2 min for a match during testing weekends.

    Look, I understand people are reluctant to change, and I don't mind explaining my thought process to people who genuinely want to discuss possibility. But this is the third time I repeat my first post through paraphrase and I'm simply too busy for that. The sheer amount of disrespect I get from people who don't even bother finish reading first post is staggering.

    What I'm suggesting is a new contingency, one with low cost but high efficiency. To put it frankly, putting the responsibility of our current balance mess to "not communicating between departments" is way too simplifying the issue. Chances are we'll be looking back at this moment and wondering whether it really repopulate the players in those game-modes, like when they excitingly explained PvP/PvE split balance few years ago.

    [RED] Crimson Sunspears...your small family guild since 2015.

  • Knox.8962Knox.8962 Member ✭✭✭

    @NICENIKESHOE.7128 said:

    @Knox.8962 said:
    You also seem to completely neglect the idea that, for PvP espcially, if you think queue times and finding groups are bad on Live servers, imagine how long they would be on the test realm if you split the playerbase between live and test.

    Hard to give feedback or test anything seriously when you don't have enough players to play wi

    And you seem to completely neglect the fact that, I say right from my first post, that this will occur in hotjoins/customized arena or EotM. There are literally no waiting time in those modes as the players participated get no reward. But beta-testing new changes itself is a great incentive for players to get into quickly and check it out, while not directly affecting the balance in live server modes like borderlands/EB and ranked/unranked. While you can say live server population are affected, I would argue that people prefer better balance in long run than having wait extra 1-2 min for a match during testing weekends.

    You seem to think that they can split skills in some way other than numerically by game subtype on live servers. They have never shown any evidence of this nor discussed it as a possibility. They have repeatedly said they can only do skill splits numerically by game mode.

    In order to test the way you are suggesting, they would have to generate new tech or create a duplicate set of test skills and traits maintain both sets on the live environment. This is most likely a prohibitively expensive option both short term and ongoing for them.

    Alternatively they could set up a test server, like they have done in the past for wvw tests and maintain a test environment that isolates balance changes from any content spoilers that they aren't ready to share. People who wanted to test would log into a different set of servers and play on characters that have no permanence and get no rewards and are effectively removed from the live games pool of available players. In my experience across several games with public test servers, they are typically ghost towns, so finding quality matches to actually test anything is usually difficult at best.

    Would a test server provide additional feedback to them? Certainly.

    Is supporting a public test environment a place I'd like to see them deploying the limited resources they have? Definitely not.

©2010–2018 ArenaNet, LLC. All rights reserved. Guild Wars, Guild Wars 2, Heart of Thorns, Guild Wars 2: Path of Fire, ArenaNet, NCSOFT, the Interlocking NC Logo, and all associated logos and designs are trademarks or registered trademarks of NCSOFT Corporation. All other trademarks are the property of their respective owners.