- Yeah, but it sounds like you're going through a list of all users and checking each one for whether they're in a specific forum group or not? Wouldn't it be better to look up the forum group and its list of members? I might be misunderstanding, but it sounds like you're doing it backwards -- going through 13,688 players to find the 10 that are in a specific forum group sounds awful.

I hope I'm just misunderstanding?

[edit]

Apparently I missed a few posts. It's possible this has already been explained. :x

[edit]

Ok, it looks like I was just confusing myself. Ignore me. :)

Okay, but still. Those 7 users? They're the ones who obviously use forum groups the most. I feel like this is the forum revamp all over again, where you revamped the forums for the people who didn't use the forums. This whole group limit thing? Is for the people who don't or barely use groups I guess. There are less than 2,000 active users a day, to be honest. 13k+ people who have joined more than 5 groups, aren't going to be coming to the forums daily, in fact, they probably don't come around the site at all, just saying. Those people shouldn't be factored in, in my opinion. You need to be more considerate of the people who actually use the feature and look at how many groups the active users are in instead, before implementing a really tiny limit based on numbers that really don't apply. :c

I might be totally wrong but it sounds like massive groups, rather than the number of groups, is the problem.

Also, I understand that 200+ groups is excessive and way out there compared to regular users but most of us posting are in way more than 15 groups. I'm in 66 and there are very few I could leave without missing out on a lot. Would a limit of say 75 be helpful at all, or is that just too high for it to make a difference?

I must agree with . I've got..what?...like 1.2k forums posts in just shy of 5 years, but I'm in 28 ping groups. I'm definitely not what you figured your average player to be with 10 to 15 minutes on site every day but neither am I the every free hour I have player anymore. (Edit: Wasn't off by much. 1.5k posts in just shy of 5 years.)

Would it not help to delete the ping groups that haven't been used in say 2 years?

If frozen accounts can be pinged (individually and/or via a group) would it not help to disable that? Or even to disable pingability for accounts that haven't shown signs of activity in a year? Mind I don't know if anything like this is even doable in the coding world.

I also agree with . It does sound like massive member lists are the issue. But again...wouldn't the problem still exist if ping groups became cults because you'd still be pinging a huge list of members.

So I understand how databases work (and that a lot of rows are a pain to work through!!), but your explanation doesn't make sense, as said (PS I do not mean I am master of all databases and don't want to sound like that I just mean like, given the knowledge I have, I'm still having some trouble understanding and would definitely love to hear/understand more)

For example, I'm in Shameless's ping group. From your explanation of table setup, the ping has to search for the forum group in the table (Shameless) then select all members of the group (me, five, far, etc etc) and ping each one individually. But that means my joining multiple groups has no adverse effect on the group in question (Shameless) - it only has the effect of making other groups n+1 members larger.

So limiting the number of groups a user can join doesn't solve the problem you're describing unless your tables aren't set up the way you just specified. All it does is ensure that at most overall to the system only (n->m) +10 (+100 for GA users) exists, but m (number of members in a group) could be anything from 0 to 6000, because that number (the one you're specifying as the load bottleneck) never was limited. You only limited the opportunity points at which m could increase, not m overall.

EDITING AS I PONDER: Also discussing the number of users to search through....if you've already entered the forumgroup table to the forumgroupusers table, all users in that table should be getting pinged. There's no need to search it. The select query might take a hefty amount of time to select and run the ping function on those 2000 users' username column, but it doesn't need to search for anyone because if the database is being kept up to date and the setup is how you seem to be describing it, then as in my example above, the Shameless table contains exactly everyone who needs and wants to be pinged by that - therefore there's no need to search users in that table, they all have to be pulled.

That leaves only the forumgroups table itself as needing to be searched for a limit 1 result (Shameless), and return its result to jump to that forumgroupusers table. So the number of overall forum groups might have to be limited, but not forumgroups a user is in.

Unless your keying for "who can ping" is the problem - as in, it has to find the user in the forum group before determining if the ping can even go through, then pulling the select; and then I see your issue, but that's a bad database setup, not a load problem (don't get me wrong I totally get setting up your database and not being able to really change it without completely messing everything up I don't mean that as in redo it at all I just meant it's not inherently load) - and depending on the setup, you could run a query for 'select can ping/permissions column, check the user id, then carry on - it would make the function 2n instead of n (having to go through once to find if you can ping, then again if you can to select all), but that's not an algorithmic difference and resolves to n time still anyway. I understand there's underlaying functions calling that are making the optimization much more intensive and the timing much higher, but I guess if we had a better understanding of how the database was set up, it would make more sense why this is such a load bearing issue?

(I'm seriously not trying to sound know-it-all I get there's a lot of complexity to subeta, I'm just trying to understand and comprehend relative to what you're explaining as the issue)

Why not actually fix your database to be able to handle the features your users who are the most active (and therefore most likely the ones buying csc to support the site) want to use instead of using the average inactive user as a guide, breaking the things your active players want. Do you really want your "po wer players" dropping to the activity level of the average user? That seems to be the road you're firmly trudging down.

I've been reading this thread over the last few days, but have wanted to stay out of it until I heard more from Keith. I agree with others that I would have liked this to be an official siteupdate announcement to begin with--that's why a lot of us are in the siteupdate group!

- I have been a user for three and a half years, I view the forums regularly--maybe not once a day, but multiple times a week--and I am in 6 ping groups, two of which are effectively defunct. I don't like that you (and others who have posted here) assume that I am not an active user. There are definitely active users who use the forums and don't have many ping groups.

I feel like a lot of the users posting here are ones who were directly affected by this change. Since it still hasn't been announced officially (eg. on siteupdate), there are probably a number of users who are unaffected and therefore completely unaware that there is a problem. I only learned about it because I regularly peruse at the site feedback topics to see what the salient issues in Subeta are.

I'm sorry that you don't like what I said. But I never said that "active" users couldn't be in only 6 ping groups. Like I said, there are less than 2,000 active users a day, and I just find it strange that Keith was using over ten thousand people who probably don't even play anymore as a factor in his averaging.

No, I don't think most people would have any reason to be in 200 groups, but 15 is such a ridiculous limit that after joining groups that staff either runs or actively recommends, you barely have any room left for other things. Maybe the average (active) user doesn't need more room, but with a massive CW market (that is a big chunk of funding for the site), that limit is way too restrictive. How is the site supposed to bring in money if they go and eliminate a crucial method of letting people know that there's a new item to spend that money on?

Omg, haha. I'm one of seven users that are in over 200 ping groups?! did you see this, we're two of the seven people! can we get a title? "destroyer of subeta" is perfect ahaha Being a ping hoarder and a group addict, I've broken subeta in so many ways until now!

Jokes aside, I don't know who the other guys are, but it's easy to determine what I and have in common. We're both stalkers? Well, that's also true xD Yet, we have a stalking foot more like both on CW/art forums. I'm not going to argue about my rights to keep every single group I've ever joined because of course I admit I join event things and never unjoin, but this might be interesting information to add: I did some counting right now and 210 out of my 248 groups are either art or CW related, out of which most are CW ones. I did not include "cw name change" and "cw announcements", since I consider that to be public utility, so that'd make it 212, give or take a few.

Of course if I'm being told this is a problem, I'm not going to question that this is a problem. As I said, I'd be willing to try new things; I just hope the above numbers illustrate that the biggest source for "group excess", at least in mine and 's case, relies on the CW market. Custom wearables' sales structure is the one in need of a "fix" if we're going to have to drop groups, which is currently tough because if we don't become stalkers, we suddenly miss everything that's happening.

[edit] Bottom line: The source of the issue isn't that users just want to join hundreds of ping groups, it's that we're pushed to do so because reasons (which are the ones that should be addressed). Just like I hoard events for ease of navigation and saving relevant site information (again, think CWs) that aren't available anywhere else on site.

I'm curious to see if there is any correlation between those that are in numerous ping groups, and the amount of money spent on the site.

Okay as I'm sitting here I actually think I understand what you mean by users bottlenecking - do forumgroups not have each their own table of data? As in, when I say forumgroup shameless, and it finds shameless in forumgroups, there is no actual table "shameless" containing all the users? Instead forumgroupusers is a list with individual lines like: user1 forumgroup1 user1 forumgroup2 user1 forumgroup3 user2 forumgroup1 etc etc etc?

So if so, the problem is that the database structure is duplicating tons and tons of data to the point that not even indexing helps, when if that data were reformatted and seperated into the tables for forumgroups as described, it would do away with that load?

I know that's not an easy fix or even potentially a viable one. It just dawned on me that I thought I might understand what you meant by users and searching clogging the system (because I didn't understand how searching in a specific forumgroup for who to ping was even important...but it makes sense in this context because you have to search by the group for each user in a listing of every group and every user multiple times).

I know a lot of us really want to understand, so it would be great if you could speak to if this is more along how the databases are currently structured. We want to understand and help if we can!

EDIT: It was really bothering me that I couldn't think of the example from my classes that I felt like fit what I was thinking the issue was so I actually dug out an old textbook until I found the term I was looking for and then found the example that helped me "get it" Fourth Normal Form and Normalization of Data Is the problem that the database is structured like the example with multiple dependencies? Think of the chains as users, the pizza types as forumgroups, and the location as ping permission/authority to ping - all of them can be keys which is the problem in structuring the data which resulted in the giant hard to search table, right? Or am I still missing the ball?

LOL YAY, one of SEVEN, said she was one too ahaha now we only need to find 4 more!

I wonder who has the most ping groups lol I'm at like 280-290 xD also I posted on keith's other board about this so gooo look there

Everyone is speaking code and math, I can't understand it! lol. My brain is like "Ow... too much thinking..." lol. At least y'all understand coding.

But I do understand one thing that has honestly confused me for sure. just said that the ping groups are formatted to search differently and separately than what the subscribed feature does. Subscribing to a thread pings you to that group when you go there, post or whatever. It says 'there are new pings in such and such thread!' when you visited a few hours ago. If you don't go visiting for a bit (I don't know what the time frame is, 24 hours?) then the pings stop until you go back to subscribed thread and check it. So it's my understanding, that this is what he is meaning to 'ease the servers' by doing double duty that since it already pings, that combining them will take a load off all the work. But I agree, I still don't understand why this is seen as the solution? It may ease the server, but annoy the userbase. Many users don't wish to be pinged to a thread every single time there is a reply. That was the benefit of ping groups and also it was a way to be more exclusive. If anyone can subscribe to a thread, that will negate all of that and while some people might be happy about that (some ppl hate private wears because they're excluded)- the sellers might find this as forcing them to conform and not be exclusive. No one can feel special, everyone has to get along with everyone else. But if you're then saying 'that's what the cult features are for' that still seems like a stretch and a more annoying method that what has been thus far.

Averaging out the entire number of users on site does not average what the actual active user is. Before I became active at all on the fourms (I didn't like just talking to people randomly for awhile)- I was still an active user because I visited the site daily, played games, sold merch, fed my pets, ect. SO there are different levels of activity, but if it's being averaged out to include over 10k user names- 80% of which are not all that active (how many users daily are onsite? 2k now? Give or take? At peak hours? Currently there are a little under 800 online. That's always been the average to me for years now, but maybe it peaks higher when I'm not online cuz I get on in the evenings my time) but the average is not averaging the number of actual active users, it's factoring ALL registered users, which does not show the correct average. The math is off. I'm not all that good at math, but at least I understand that much...

The one question I have too... is if you're saying 'out of 13k registered users' that have joined ping groups, are you saying that the numbers of those that login daily and use the site, let's say 1000 (just throwing out a number) that that number is different users every day? So every day over the course of 24 hours, all 13k members login and use subeta at some point? So you can say- only about 100 users are the same users on at the same time, but 12,900 other members are on throughout the day? Every day? Using these features you need to keep the site up and running? Or, is it that let's say, 10k other registered users only get on to check Subeta once every month. Or, is it, that say half get on about once a month for about an hour-and the other half are completely inactive? Either way, the averaging still doesn't add up to what and who is actually using the features available in the forums actively. Two user may have only joined 5 groups each, but one actually uses Subeta while the other doesn't. The count needs to take into consideration who is actually using said features and who will be benefited by any changes.

We will all be affected. I get that. I also understand needing to put limits on things in order to run things smoothly. But the other thing to take into consideration, is if you are ignoring the said "power users" and just lumping them in with other users that are active and use the forums, but don't user the Commerce section- you are essentially saying every user is the same which they are not.

Perhaps you ought to look into finding a way of rewarding 'power users'. You should be able to track who the biggest contributors to the site are, esp in regards to cws. Those that have released the most should perhaps have privileges so they don't lose the convenience of selling on subeta. The change will affect the commerce part of the site quite a bit by limiting the ability to sell and how selling works. Many people purchase things on site with csc which they spend their real money on so they can buy stuff. The CW market has brought in a lot of money, given exposure to many artists who have in turn also been able to benefit from selling their work in the art selling forum for csc as well. Overlay artists, drawing OCs and ect. People creating giveaways, contests, auctions ect won't be able to do it the same anymore because the limit places an awkward restriction on them.

Like others have already stated, there just seems that there could be a better way to set things up instead of merging everything together and hoping for the best.

I've rambled too much... lol.

After counting the groups in the "your groups" section for pings... I can say that I guess I'm one of the seven people that are in more than 200 groups?

Both happy and "whoops" at the same time.

Unless the "your groups" for ping groups double up like the achievement listings ? Since I just counted as being in 650+ groups? (that cant really be right?)

That's what I was asking if that was the case. The problem in my opinion is, is the fact that you don't have a definite answer about who is online, just that it's 'most likely the case' and honestly, I don't think making an average on probably's is good enough. There should be harder numbers to make a more accurate count. Plus, even so, why would you change the system to work for the 'average' user that doesn't spend enough time online in the first place to be of any benefit? My assumption is your hope is that these changes will entice these users to spend MORE time on site. But chances are, by definition, they won't and such drastic changes w/out more benefit to the user that spends more time and money on site will only alienate them and instead of seeing more use on site, you will see less. (Or, it will take longer for the 'average' user to get using the site to the point to make up the difference of the loss and therefore your profit will go down and take longer to recover.)

Just sayin'. More definite numbers and information would be of a better benefit to ensure certain changes are what is best for the site as a whole and to ensure the best profitability for your site. I just see this change as hurtful for the profit line, not helpful. You may say that in the long run it'll be better, but I'm not sure taking such a risk that may end up dipper farther than you have the time (and money) to wait for it to get more level and then continue to rise. If I'm even making sense.... I don't know anymore. lol.

I understand what you mean though, but still, I don't think making decisions on the 'chance' you are correct is a very smart or business savvy method.

I counted my groups. I'm with you and Clamp - well over 200 here. Most of them are active as far as I can tell.