Although we've had a long crazy trip here, we're back to square one.

The new population caps throw off all the math. I can't do this anymore until and unless I get a lot more data on what server status = how many people.

If you're interested in helping, please take this survey HERE a few times a week, on the imperial and republic sides of your server, and at different server loads (light, medium, heavy, very heavy, full).

The data, as always, will be housed HERE so that everyone can see it and do their own analysis if they choose.

Paige

Updated spreadsheet is here - this is where the good stuff lives.

https://docs.google.com/spreadsheet/ccc?key=0Aregkvys5QFodFJ2OWN5U0hwaVFBYWdqUUh1WmdZUFE#gid=31

New numbers are up and exciting! We've got a continued climb in population... we're up to 66,996 estimated concurrent logins.

Date Estimated concurrent logins

5/8/2012 100,002

5/11/2012 95,859

5/14/2012 95,946

5/16/2012 85,995

5/18/2012 83,194

5/21/2012 76,255

5/23/2012 73,693

5/25/2012 72,203

5/28/2012 66,574

5/30/2012 66,323

6/1/2012 65,772

6/5/2012 63,243

6/6/2012 62,891

6/8/2012 62,601

6/11/2012 62,431

6/13/2012 62,923

6/18/2012 66,996

Again looking at things by transfer status gives us some interesting comparisons.

SUM LOGINS BY Transfer Status

Destination 16,038

Origin 37,558

Origin Likely 10,565

Unknown 2,835

Total 66,996

PER SERVER AVG LOGINS BY Transfer Status

Destination 802

Origin 249

Origin Likely 258

Unknown 473

Grand Avg 307

The population of the origin servers is still looking high, but that's diluted out in the 2 week average. The Destination servers have a whopping 800ish average logins per server! That's the highest number I've seen since starting this process.

Paige

Updated spreadsheet is here - this is where the good stuff lives.

 

https://docs.google.com/spreadsheet/ccc?key=0Aregkvys5QFodFJ2OWN5U0hwaVFBYWdqUUh1WmdZUFE#gid=31

 

New numbers are up and exciting! We've got a continued climb in population... we're up to 66,996 estimated concurrent logins.

 

Date Estimated concurrent logins

5/8/2012 100,002

5/11/2012 95,859

5/14/2012 95,946

5/16/2012 85,995

5/18/2012 83,194

5/21/2012 76,255

5/23/2012 73,693

5/25/2012 72,203

5/28/2012 66,574

5/30/2012 66,323

6/1/2012 65,772

6/5/2012 63,243

6/6/2012 62,891

6/8/2012 62,601

6/11/2012 62,431

6/13/2012 62,923

6/18/2012 66,996

 

 

Again looking at things by transfer status gives us some interesting comparisons.

 

SUM LOGINS BY Transfer Status

Destination 16,038

Origin 37,558

Origin Likely 10,565

Unknown 2,835

Total 66,996

 

PER SERVER AVG LOGINS BY Transfer Status

Destination 802

Origin 249

Origin Likely 258

Unknown 473

Grand Avg 307

 

 

The population of the origin servers is still looking high, but that's diluted out in the 2 week average. The Destination servers have a whopping 800ish average logins per server! That's the highest number I've seen since starting this process.

 

Paige

It is a increase of logins over what it was a week or so ago for sure. But it is nowhere near close to what it was back last month.

Thanks OP. Some interesting data. What and can we gather from this which may show what the total sub base is for TOR now? Or is that info still subjective at best?

Thank you. That's a great question, and I know of no way to answer it (though I'm happy for someone else to show me the way.) The big sticky wicket is that I can estimate the number of people logged in but I have no clue who these people are. It's like the example with the housewives, school kids, salarymen, college kids, insomniacs, and 3rd shifters who all log in at different times. I don't have any good way to understand what percentage of the population is actually logging in and playing. I'm a GL of a casual guild, and there are plenty of people in my guild who are completely satisfied to pay $15 a month to login and play for 4 hours one night a week. We also have a couple of very-determined stay-at-home parents, a retiree, and a couple of college professors with not much to do this summer who play for 6+ hours a day. Time in game is wildly variable in my own guild and circle of friends, I imagine that it's wildly variable in the general population, too.

So, I'm sorry, but I got nothing.

BW has said 1.3 m subs. Assuming that's still true (which is an extremely uncertain assumption) then that says about 5% of the population is active at any one time. I guess that doesn't sound unreasonable to me, but I have no solid quantitative data to back up that "feeling".

Paige

Thank you. That's a great question, and I know of no way to answer it (though I'm happy for someone else to show me the way.) The big sticky wicket is that I can estimate the number of people logged in but I have no clue who these people are. It's like the example with the housewives, school kids, salarymen, college kids, insomniacs, and 3rd shifters who all log in at different times. I don't have any good way to understand what percentage of the population is actually logging in and playing. I'm a GL of a casual guild, and there are plenty of people in my guild who are completely satisfied to pay $15 a month to login and play for 4 hours one night a week. We also have a couple of very-determined stay-at-home parents, a retiree, and a couple of college professors with not much to do this summer who play for 6+ hours a day. Time in game is wildly variable in my own guild and circle of friends, I imagine that it's wildly variable in the general population, too.

 

So, I'm sorry, but I got nothing.

 

BW has said 1.3 m subs. Assuming that's still true (which is an extremely uncertain assumption) then that says about 5% of the population is active at any one time. I guess that doesn't sound unreasonable to me, but I have no solid quantitative data to back up that "feeling".

 

 

Paige

Forget the outliers and go by the 10% rule that seems to be acceptable by MMO developers and designers--

"On an average 10-15% of the subscribed player base is online at a time in an MMO"

Forget the outliers and go by the 10% rule that seems to be acceptable by MMO developers and designers

That sounds as reasonable as anything else. :-) Given that this is the internets and all, do you have a citation for that so I may review it?

Paige

I just measured a few servers post transfer using the /who 'level range' method to compare against the assumption

Light is up to about 500

Standard is up to about 1,500

Heavy is up to about 2,250-2,500(?) (Let's call this 2375 for now...)

Very Heavy is up to about 3,000

Full is greater than about 3,000

I think that the Light, Standard and Heavy are still as suggested.

Full on the other hand is suspect and the one he had least data on, my one and only Full server measurement threw up a number of c3800.

There are two possibilities;

1. They have increased the 'Full' limit from c3000 to c3750.

2. 'Full' itself has a range e.g c3000-c3750. The reason I suggest this is that the server in question (TOFN) remained at 'Full' for some time (minutes not hours) before indicating there was a queue and therefore the initial assumption was always wrong.

And in the time it took to write this the queue has increased from 5-20 minutes and the actual queue is 250 meaning that I'm measuring at peak log in time.

nice work

I agree NICE WORK

I just measured a few servers post transfer using the /who 'level range' method to compare against the assumption [snip]

I think that the Light, Standard and Heavy are still as suggested.

Full on the other hand is suspect and the one he had least data on, my one and only Full server measurement threw up a number of c3800.

 

There are two possibilities;

 

1. They have increased the 'Full' limit from c3000 to c3750.

2. 'Full' itself has a range e.g c3000-c3750. The reason I suggest this is that the server in question (TOFN) remained at 'Full' for some time (minutes not hours) before indicating there was a queue and therefore the initial assumption was always wrong.

 

And in the time it took to write this the queue has increased from 5-20 minutes and the actual queue is 250 meaning that I'm measuring at peak log in time.

Hunh, interesting. If you want, you're welcome to enter the info here.

http://www.thethirteenthlegion.torportal.com click survey on the upper right.

That will warehouse the info in a publicly available place so anyone can do math at it.

I'm running a similar experiment with the Ebon Hawk. I ask people to /who the server during peak hours and report back the status.

Average of Assumed Total Pop

Standard 2088

Heavy 2216

Very Heavy 3419

Full 3166

Very heavy > Full. I know. I know. It's what the math says though. More data will hopefully sort that out.

Here's the data.

Status Total Pop (2x faction pop)

Heavy 1740

Standard 2088

Heavy 2242

Heavy 2242

Heavy 2338

Heavy 2516

Very Heavy 2844

Full 3166

Very Heavy 3994

Paige

Nice work but.... why?

I don't get why anyone but Bioware or their competition would care.

A) Because I can. I'm doing it because I'm curious.

B) I want to see where the population shifts to so that I can have some ability to predict if an equally large upset will occur again so I can prepare ahead of time.

Paige

So this is one week before transfers began, and one week after. The Jung Ma number seems like a before transfers number. There's been a noticeable spike since transfers.

WIll look forward to the averages that *don't* include pre-transfer.

I, too, just find it interesting.

Plus, I find it nice to see the increase in population [in Begeren Colony]. I can do a /who of 1-15 now, and get over 100 people! Before (when I was just checking numbers, not entering on the form) I'd be able to do 3 searches ( 1-25, 26-49, 50 ) and get the total population on one side.

I'll be interested to see if we continue to grow as more people transfer over, or if the rest are staying at their original server and we even out.

p.s. Great work, Paige!!

Hi OP:

This is great work, thank you.

If you can keep running these numbers, you'll see trends as the subscribers trend UP or DOWN in the coming months.

[snip] There's been a noticeable spike since transfers. WIll look forward to the averages that *don't* include pre-transfer.

Holy catfish yes! That will be very interesting.

I'll be interested to see if we continue to grow as more people transfer over, or if the rest are staying at their original server and we even out.

 

p.s. Great work, Paige!!

I know, right? Why would anyone stay? Is anyone staying? Does it look like it would make sense to shut down the origin servers? Can't wait to see more data, and thank you for your kind words.

Just a fyi, I have been away the last wek at the leman 24h, got back today and logged into my origination server (Hexdroid) and I was the only player online in either faction when I logged them in to clear their mailboxes. So with that as an example of an Origin server population I think you can count them all as 0 pretty much.

I wondered why I hadn't seen you posting! I should go back and check my old server too. I think once we get in to entirely post-transfer data, we'll see all of the origin servers at or near a phi of 1...

This is great work, thank you.

If you can keep running these numbers, you'll see trends as the subscribers trend UP or DOWN in the coming months.

Thank you! I do intend to keep running them MWF for the forseeable future. :-) I suspect that after the excitement of mergers wears off, then we'll see a drop in logins to some "normal" level. Just remember, we can't really speak to subscriber numbers. It's like the example of the stay-at-home parents, school kids, salarymen, college students, insomniacs, and 3rd shifters all logging in at different times. We don't know anything about subscriber numbers.

Cool stuff!

Thanks!

Paige

TL;DR This post is intended for Scorpienne/Paige and contains a lot of math. To anyone else who reads and appreciates it thank you.

Paige, first let me start by saying thank you for the hard work you have done over the last several weeks compiling numbers and posting them to the boards. They have definitely been an interesting read and provide a good insight into the number of players SWTOR has. However I am here because because I feel your numbers are wrong.

I think the reason you haven't caught the error yet is because most of the sever pops were too low, but now with the transfers I wanted to let you know about the mistake and show you a new formula before numbers got out of control.

First let me state why I feel your numbers are wrong. To use your example of The Fatman (currently 2.52). If that number were say 3.20 (a very reasonable number with the increased players from the transfers) based on your current formula it would have an average population of 250 + 1000 + 1938 + (.2)2687.5 = 3725. That means a server fluctuating between heavy and very heavy would have an average population over full capacity? That doesn't make much sense.

Because this seemed off to me I started to think about what the torstatus numbers meant and how they could be used to find the average population using a different formula. I began to look at the status as a % of time at a certain population. For example to go back to The Fatman and 2.52. to me that means the server was at standard (2) 48% of the time and heavy (3) 52% of the time.

2(.48)+3(.52)=2.52

as proof that this works for all statuses and not just the one i picked i offer this generic variable equation.

torstatus = x+(n/100) where x = whole number from 0-5 and n= whole number 0-99

y= x+1

x+(n/100) = x[(100-n)/100] + y(n/100)

x+(n/100) = [(100x-xn)/100] + [(n+nx)/100]

x+(n/100) = x+(n/100)

(yes i skipped steps but its algebra, feel free to write it all out if you don't believe me)

Since it is easy to substitute a population number for the torstatus number such as 2 = standard = 500-1500 I would content that the average population should be taken from those %

To continue using your example of the Fatman lets say 48% of the time it is standard, and 52% of the time its heavy. That means the fewest amount of people would be 500(.48) + 1500(.52) = 1020, (lowest values of standard and heavy) and the highest would be 1500(.48) +2375(.52) = 1955 (highest values of stand and heavy) this would mean the average population at any given time should be somewhere in the middle. if you add those up and divide by 2 you come out with 1488. To simplify this you can use your already established center numbers for each bracket (250, 1000, 1937.5, etc) instead of doing the math twice.

Here is the proof for that in case anyone has questions

x= %1, y=%2, a = lowest pop 1, b = lowest pop 2, c= highest pop 1, d= highest pop 2

{[(x(a)+y(b)]+[x©+y(d)]/2} = x[(a+c)/2] + y[(b+d)/2]

where [(a+c)/2] and [(b+d)/2] = the already established center numbers

So to summarize I think you can get a much more accurate number for the average population by taking the torstatus number and using it to determine the % of time a server spends at the 2 surrounding statuses. So for a final example I will again use Fatman at 2.52. Take the % at standard (48) and the % at heavy (52) and then replace the standard and heavy with the established numbers for those designations.

1000(.48)+1937.5(.52) = 1488

While this is much lower than your current estimation I feel it is much more accurate. And it makes sense if you think about it. A server that spends about half its time at standard and half at heavy is probably always right around that threshold for the 2 designations, in this case 1500.

If you have any questions, or need anything explained in more detail I will be glad to reply, or if you want to disprove my theory I welcome that as well.

Thanks for your time!

Jeff

Holy catfish .....

 

I know, right? Why would anyone stay? Is anyone staying? Does it look like it would make sense to shut down the origin servers? Can't wait to see more data, and thank you for your kind words..

 

 

Paige

Paige..Paige... you seem to be caught up in the thrill of the moment here.

Why? You ask... because if I accept the so-called transfer "choice", I will have to delete 4 of my characters, and have now a totally full 8 character server with no room for new species, or legacy species.

So I log-in stare at servers, like a deer in the headlights of a semi-truck... and then log-out.

TL;DR This post is intended for Scorpienne/Paige and contains a lot of math.

Whoa! I think I have met a kindred spirit! I'm always up for math, especially when it's presented beautifully and politely. Thank you.

That is an awesome way to look at the numbers and one I hadn't thought of at all. Makes good sense, because the category **includes** the categories below it. No sense in counting the midpoint of the light status, if you know that heavy means approximately X many logins.

Hm... on the low phi servers though... using the Jeff Method a status of 0.99 means it spends 1% of the time at 0 and 99% of the time at 1. Assuming status 1 ≤ 500 people that means our low-phi servers pop levels rise dramatically from an estimate of 250ish to an estimate of nearly 500ish. These low phi servers are a pain in my broad pink bottom because they just do *not* have enough information. The detector (torstatus) just doesn't work at that low level. I hate to just throw my hands up in the air and say it's too uncertain to calculate... but I'm really rapidly reaching that conclusion. I think that I'll just have to lump any server with a phi of ≤ 1 as a group and say it's ≤500 people and so it could be from 0 to 500x the number of servers logins and accept it as an uncertainty in the analysis.

Lemme go play with some spreadsheets and I'll be back.

Paige

 

These low phi servers are a pain in my broad pink bottom because they just do *not* have enough information. The detector (torstatus) just doesn't work at that low level. I hate to just throw my hands up in the air and say it's too uncertain to calculate... but I'm really rapidly reaching that conclusion.

 

Yes the low pop servers are almost impossible since if a server never goes above the light threshold there is no way to know if there are 5 people or 499 people. I do believe though that our 2 methods would come out to the same average population for the light servers though. So if you stick to calling them all ~250 I think it would work, since some would be lower and some higher in reality.

The real problem now though is with transfers, most of the origin servers are going to fall into a perpetual state of light status, while the destinations servers are going to spike to much higher numbers. There is no way to know if the light origin servers are barren, or if enough people stayed to make them viable for evaluation.