Suggestion - How about shorter betas?

[b w.7389]

It looks like this week is going to go a lot like previous beta weeks -- some problems are fixed, and there are some new ones.

And, just like previous weeks, these issues have become apparent really quickly. For instance, a lot of us are frustrated because we're not able to play with our friends or family because of the assignment issues.

 

Instead of making us cope with the same issues all week, how about a 24 hour beta test and then switching back to normal? There's really no advantage in running it longer (we're not testing any "end of week re-matchmaking" or anything).

[Drendal.2085]

shorter would be great. i only have the next two weeks to play before im deployed and would like to see my friends in game.

[Arhes Branwin.2095]

No kidding, I'd like to play with friends, this is an MMO after all.  Just end it and re-do it when you can actually get people together in the right places.  Otherwise, you've made a mode worthless to me for a week.

[subversiontwo.7501]

I'd assume they're just running it a full week because some players enjoy the fresh feel to it, like any event. It's not like running multiple smaller resets allow them to sift through the data and program adjustments quicker anyway. The development cycles on this likely run on months rather than weeks. Week is just a familiar runtime for players. It may ramp up as they get closer to solving the issues presenting themselves in the current step.

This first step of the beta will keep running until they get to the bottom of the problems I assume. Then there are the algorithms to test, then there is the actual Alliance function to test and then there are the cap- and transfer functions to test as well as whatever matchup and ladder replacement to develop and test too.

If anything, that is the only somewhat alarming bit, that things still keep progressing this slow this many years into it and these many months after the last declaration of going into beta. This is beta 3 but it is still essentially beta 1. They really need to slay these bugs now to keep working on the stuff in any sense of continued timely fashion, even if what they work on now may be the most fundamental and difficult to get stable enough in a programming sense.

Edited January 15, 2022 by subversiontwo.7501

[Dawdler.8521]

I would rather say take a week and reset like normal next friday, but during that week reset like friday, tuesday and thursday in order to squeeze in more team assignment testing. But I'm not really sure they can do fixes/tweaks that fast.

[Custodio.6134]

the thing is: if you want to get accurate data, you need a proper amount of data. It´s called "sample size". 
The less time you run a beta, the less data you get, and the more inaccurate the data becomes. I mean: directly at reset the activity is probably the highest you will ever get, and over the course of the week that changes (also because of weekend). you also cannot collect daytime-based data just within some hours, and you will never get data based on certain days/parts of days (morning/afternoon/evening/night). You also cannot account for timezones if you run the beta for just a few hours. It would make the whole point of testing invalid. 

It would be even better to make the beta-periods longer, as you increase the sample size and it would consider different matchups in the datasets, but people will complain even more. So the best choice for ANet is actually a full matchup (aka 1 week) at the moment. 

 

3 hours ago, subversiontwo.7501 said:

This first step of the beta will keep running until they get to the bottom of the problems I assume. Then there are the algorithms to test, then there is the actual Alliance function to test and then there are the cap- and transfer functions to test as well as whatever matchup and ladder replacement to develop and test too.

If anything, that is the only somewhat alarming bit, that things still keep progressing this slow this many years into it and these many months after the last declaration of going into beta. This is beta 3 but it is still essentially beta 1. They really need to slay these bugs now to keep working on the stuff in any sense of continued timely fashion, even if what they work on now may be the most fundamental and difficult to get stable enough in a programming sense.

exactly. We are currently in the earliest stage of development that ANet is able to release to the players for testing. And it´s also one of the most important parts of the full system that will replace servers. They need to make sure that the core system works as good as it is able to, in oder to prevent problems that may occur in future versions (respectively additions to it, like alliance-functionality). 

[tails.4209]

It's clear most of the issues happen around reset when players are essentially DDoS'ing WvW. Presumably there's some issues in scaling the code that Anet's internal QA is not catching that is caught when run through the entire player population. To collect more data on these issues, they should have more frequent resets. The quality of gameplay over the term of a week is not being tested during these betas right? Even if it is, it should be secondary to fixing the distribution of teams. Having players be put on the right team is the objective of alliances after all. I would consider bugs like the wrong team one a showstopper.

[Zok.4956]

On 1/15/2022 at 7:28 PM, b w.7389 said:

Instead of making us cope with the same issues all week, how about a 24 hour beta test and then switching back to normal? There's really no advantage in running it longer (we're not testing any "end of week re-matchmaking" or anything).

Yes, shorter cycles would be good for players, but I guess 24hours is too short.

Errors have to be reported and then the devs have to look into the actual data and try to find the bugs and fix them. This does take some time, depending on the bug. But if a "reset" is too fast, the real production enviroment transaction data is lost and it probably can not be reproduced with real-data backup 100% in a testing environment. And if the bugs are not fixed they will happen again after the "reset". Nothing won. It is a legacy software system with probably a lot of unexpected internal dependencies.

Edited January 17, 2022 by Zok.4956

[The Boz.2038]

"I joined the one-day beta on the tail end, my team got beaten, 1/10, system not working as intended."