Homecoming player count: a year in review

[macskull]

"Homecoming is dying."

 

"The servers are empty, no one plays anymore."

 

"It's so hard to find teams even on Excelsior."

 

"The lower-population servers are a ghost town."

 

"Fire Blast is in desparate need of buffs."

 

"Regen is a good set when you use an entire tray of oranges and purples every fight."

 

Have you or a friend heard someone utter those phrases, or ones like them, ingame or on the forums? Or even on Discord? Well, here's some cold hard data.

 

Background:

 

Last March I was curious about player count because it felt like less people were playing, but I had no actual data to back up my feelings. Sure, the HC Discord has player count statistics, but they only go back seven days so they're kind of useless for long-term analysis. I initially started my tracking by manually checking player counts from the server status page at a certain time once a week. This was fine, unless I didn't have an internet connection at 6PM on a Saturday, or I forgot to check, or I forgot to write down the number, or Mercury was in tardigrade, or Corey was in the house, or whatever else the cosmos could dream up to disrupt my plans. Oh, and I was about to drop off the face of the earth for half a year with no way to even get onto the internet, let alone update anything at all, and a tracking system like this that relies on manual labor is definitely not going to work when the person doing that labor isn't around to do it. "Surely there must be a better way," I thought.

 

Turns out, there is.

 

Enter Python and my "the only coding experience I have is a few months of Java 15 years ago and even then I could barely make a functional program" self. Turns out, you don't really have to write your own code from scratch when what you need is 90% complete somewhere else - you just have to find the pieces you want and put them together, then make them work for your specific task. In my case, I wanted a script that would do the following things:

1. Pull the player numbers from the server status page.
2. Put those numbers into a Google Sheets document.
3. Repeat steps 1 and 2 at a certain interval.
4. Use Google Sheets to determine the maximum player count for each server (and total) for a given day.
5. Graph these maximums.

 

Simple, yeah? Not so much. So many headaches. Took me a solid 3 days of tweaking code and tweaking my Sheets document to get things working. Google Sheets is weird about filters and zeroes, so I had to get creative with Google's query language to get the data in a format that would give a user-friendly graph.

 

The result:

 

A 50ish-line Python script which does everything in steps 1 and 2 above, an AWS virtual host which handles running the script every 15 minutes (and hosts everything since my computer is not always online), and a Google Sheets document with... just a few lines of data.

 

 

Today's the one-year anniversary of when I got the whole thing up and running, so here's a year in review!

 

So, player numbers did what, exactly?

 

 

This. They did this. (Ignore those weird dips in January and February, the script pulled no data for a week or so for some unknown reason). There are some significant dates for the peaks and valleys:

1. 28 August 2022: Issue 27 Page 4 released. 2378 peak players, almost 200 more than any other day over the past year.
2. 25 December 2022: Christmas. 1450 peak players. Makes sense, right?
3. The release date for Issue 27 Page 5 isn't statistically significant and is buried somewhere in those peaks in mid-October 2022.

 

More macro analysis: player numbers are roughly similar to where they were this time a year ago. They peak in the summer then fall off in November and December.

 

TL;DR: Reports of Homecoming's death are greatly exaggerated. That being said, there is extreme stratification in the server populations. Excelsior tends to have a population greater than the other four servers combined and Indomitable and Reunion really are mostly empty. I suspect there is nothing that can be done about this, but player population and proportional share of server population has remained relatively stable so I suspect we're simply sitting at a baseline. In other words, the people who are still playing the game four years after it came back into the public eye are here to stay, and any large population changes (down or up) aren't going to happen.

 

One more thing: If you want to play around with an interactive version of the graph above, check the links in my signature.

Edited April 28, 2023 by macskull
Added a wee bit

[InvaderStych]

Thanks for putting this together!

 

Also, when was Mercury in tardigrade? The timing might explain some things for me about the past 18 months. 🤣

[TheZag]

I would have expected a graph over a year to be less jagged.  Is that weekdays compared to weekends?  

 

[Snarky]

Tony Stark had dancing girls when he made his presentation.  Just sayin'...

[macskull]

22 minutes ago, TheZag said:

I would have expected a graph over a year to be less jagged.  Is that weekdays compared to weekends?  

 

Weekends are the peaks. If you look at the graphs on Discord you can see the same thing happen.

[Judasace]

1 hour ago, macskull said:

"Regen is a good set when you use an entire tray of oranges and purples every fight."

 

You forgot that you also need reds, yellows and then 1 blue at the end.

Edited April 28, 2023 by Judasace

[Veracor]

Nice!  I like posts with hard data.  It has certainly felt to me that for the last two years there's been a constant stream of new globals showing up and a similar stream of older globals no longer showing up, as players come and go.  But it has never felt like one was greater than the other, excepting events.

 

Good to have a data-driven response though for when some negative nancy says that "Homecoming is dying," and I anecdotally feel that the population is stable.

[FupDup]

6 minutes ago, Judasace said:

 

You forgot that you also need reds, yellows and then 1 blue at the end.

And Rune of Protection.

 

And Shadow Meld. 

[DoctorDitko]

Being picky:

Mercury goes into retrograde (appearing to reverse direction in the sky), not tardigrade (Adorable microscopic waterbear critters than can survive in space).

 

It won't fit in a tardigrade. I know I'm being "pendantic" (cf SGU*) but it itches!

 

* Skeptic's Guide to the Universe. World's greatest podcast: theskepticsguide.org

[MyriVerse]

1 hour ago, macskull said:

Mercury was in tardigrade

Kinky.

[SeraphimKensai]

1 hour ago, macskull said:

"Homecoming is dying."

 

"The servers are empty, no one plays anymore."

 

"It's so hard to find teams even on Excelsior."

 

"The lower-population servers are a ghost town."

 

"Fire Blast is in desparate need of buffs."

 

"Regen is a good set when you use an entire tray of oranges and purples every fight."

 

Have you or a friend heard someone utter those phrases, or ones like them, ingame or on the forums? Or even on Discord? Well, here's some cold hard data.

 

Background:

 

Last March I was curious about player count because it felt like less people were playing, but I had no actual data to back up my feelings. Sure, the HC Discord has player count statistics, but they only go back seven days so they're kind of useless for long-term analysis. I initially started my tracking by manually checking player counts from the server status page at a certain time once a week. This was fine, unless I didn't have an internet connection at 6PM on a Saturday, or I forgot to check, or I forgot to write down the number, or Mercury was in tardigrade, or Corey was in the house, or whatever else the cosmos could dream up to disrupt my plans. Oh, and I was about to drop off the face of the earth for half a year with no way to even get onto the internet, let alone update anything at all, and a tracking system like this that relies on manual labor is definitely not going to work when the person doing that labor isn't around to do it. "Surely there must be a better way," I thought.

 

Turns out, there is.

 

Enter Python and my "the only coding experience I have is a few months of Java 15 years ago and even then I could barely make a functional program" self. Turns out, you don't really have to write your own code from scratch when what you need is 90% complete somewhere else - you just have to find the pieces you want and put them together, then make them work for your specific task. In my case, I wanted a script that would do the following things:

1. Pull the player numbers from the server status page.
2. Put those numbers into a Google Sheets document.
3. Repeat steps 1 and 2 at a certain interval.
4. Use Google Sheets to determine the maximum player count for each server (and total) for a given day.
5. Graph these maximums.

 

Simple, yeah? Not so much. So many headaches. Took me a solid 3 days of tweaking code and tweaking my Sheets document to get things working. Google Sheets is weird about filters and zeroes, so I had to get creative with Google's query language to get the data in a format that would give a user-friendly graph.

 

The result:

 

A 50ish-line Python script which does everything in steps 1 and 2 above, an AWS virtual host which handles running the script every 15 minutes (and hosts everything since my computer is not always online), and a Google Sheets document with... just a few lines of data.

 

 

Today's the one-year anniversary of when I got the whole thing up and running, so here's a year in review!

 

So, player numbers did what, exactly?

 

 

This. They did this. (Ignore those weird dips in January and February, the script pulled no data for a week or so for some unknown reason). There are some significant dates for the peaks and valleys:

1. 28 August 2022: Issue 27 Page 4 released. 2378 peak players, almost 200 more than any other day over the past year.
2. 25 December 2022: Christmas. 1450 peak players. Makes sense, right?
3. The release date for Issue 27 Page 5 isn't statistically significant and is buried somewhere in those peaks in mid-October 2022.

 

More macro analysis: player numbers are roughly similar to where they were this time a year ago. They peak in the summer then fall off in November and December.

 

TL;DR: Reports of Homecoming's death are greatly exaggerated. That being said, there is extreme stratification in the server populations. Excelsior tends to have a population greater than the other four servers combined and Indomitable and Reunion really are mostly empty. I suspect there is nothing that can be done about this, but player population and proportional share of server population has remained relatively stable so I suspect we're simply sitting at a baseline. In other words, the people who are still playing the game four years after it came back into the public eye are here to stay, and any large population changes (down or up) aren't going to happen.

 

One more thing: If you want to play around with an interactive version of the graph above, check the links in my signature.

This is amazing. Good work Mac. I love me some statistics.

[Judasace]

Also  having hopped around and tried some other servers recently to see what they're doing (Some of it is cool, some of it sucks and some of it is straight up insane) I can say, if I'm reading the graphs right, that pretty much every other free server struggles to even reach 50 players at their peak times, so even though  Torch, Indom and Reunion are extremely low, their populations are probably as much or more than any other CoH server out there. Ev and Ex dwarf all of them combined by orders of magnitude, obviously.

 

I say this not to create server drama, but to point out to anyone that thinks those three servers should be axed and incorporated into the other two, that there are numerous people happily playing away on servers with that population size or smaller.

[Fire Chief]

Sturgill Simpson approves of these stats.

[Number Six]

That looks accurate and matches up with our internal stats (it should since they're both pulling from the same API). The weird looking gap in December is because either the API endpoint or the stats database was down for a few days over Christmas, but was low priority enough that nobody took time away to go fix it.

 

[twozerofoxtrot]

3 hours ago, macskull said:

"Fire Blast is in desparate need of buffs."

 

Luckily this critical shortfall has been addressed in Page 6.

 

Rest easy.

[Gulbasaur]

7 hours ago, macskull said:

"The servers are empty, no one plays anymore."

I have noticed a bit of  a "we don't use the LFG channel, join our Discord to find out blah" trend , so it feels more like a ghost town than it is at times. If you put it on the LFG, people will come. Sometimes, you have to take the lead. I'm in the UK and play on Everlasting because from experience RP servers generally have a nice atmosphere and this game ran playably on dial-up so lag is rarely a problem. 

 

A similar thing happened on the NA GW2 servers but not the EU servers (well, to nearly the same extent) where the NA LFG is a ghost town because the LFG peand the EU LFG is live and well as long at it's not like 3am on a Wednesday. 

Edited April 28, 2023 by Gulbasaur

[TheZag]

10 hours ago, Number Six said:

That looks accurate and matches up with our internal stats (it should since they're both pulling from the same API). The weird looking gap in December is because either the API endpoint or the stats database was down for a few days over Christmas, but was low priority enough that nobody took time away to go fix it.

 

 

I was able to recover the data from the missing portion of the graph

 

 

[Yomo Kimyata]

@macskull, thanks for putting this together.  It corroborates some of my observations.  I strongly suspect that the number of active users is less than that, since there still are a bunch of dual/triple boxers out there, but my estimation of 1-2k consistent active users feels pretty solid.  It's a village, and I think a high percentage of that village uses these forums.

 

And we all know Excelsior is too crowded; no one goes there anymore!

[Uun]

I'm surprised there's not a downward slope on Torch. I've been spending more of my time on Excel (and moving characters from Torch to Excel) due to the low population and many of my friends have been doing the same. 

[Clave Dark 5]

2 minutes ago, Uun said:

I'm surprised there's not a downward slope on Torch. I've been spending more of my time on Excel (and moving characters from Torch to Excel) due to the low population and many of my friends have been doing the same. 

The characters I liked I moved ages ago, maybe that's been common too.

[Scarlet Shocker]

13 hours ago, DoctorDitko said:

It won't fit in a tardigrade. I know I'm being "pendantic" (cf SGU*) but it itches!

 

 

I have it on good authority that Tardigrades are bigger on the inside than the outside

[Yaliw]

Nice squiggles! Regen for tankers please.

[Excraft]

Is this counting real players or just accounts?  What portion of these are people double or triple boxing on multiple accounts?

[Doomguide2005]

47 minutes ago, Excraft said:

Is this counting real players or just accounts?  What portion of these are people double or triple boxing on multiple accounts?

     Does it matter it terms of active players?  Even if it's all 3 boxers, those total numbers are fairly stable over the course of a year.  It would be interesting to know perhaps if it went from all 3 boxers to single player accounts over the year versus i.e. if the proportion of multi-boxing to single players was also stable or not but in terms of HC dying or not not so much I'm thinking.

[macskull]

1 hour ago, Excraft said:

Is this counting real players or just accounts?  What portion of these are people double or triple boxing on multiple accounts?

It isn’t possible to count unique players without IP address data, and I’m not certain even the devs have that level of fidelity.

 

However, we can make some assumptions and observations and infer an answer. Player counts are stable over the last year, and the individual server populations have remained proportional to the total population. This could mean a lot of players multibox a lot of the time, or it’s relatively uncommon, or - most likely - it’s somewhere in the middle. There is no reason to think the proportion of players multiboxing changes enough to matter, or we would see that reflected in the graph.

 

Regardless of multiboxing, the point of this isn’t to focus on the number of unique players but rather to show that the total population has remained relatively constant.