Weekly Update Trends (2014 - 2019) Part 1: Player Numbers
Jan 3, 2020 13:24:35 GMT -5
LordOfChange, tedium, and 2 more like this
Post by mehtastic on Jan 3, 2020 13:24:35 GMT -5
Introduction
If you care or have cared at all about Armageddon in the past, you’ve probably thought about the game’s current state and what things might look like in the future. Well, I did, at least. And so did my friend, the former staff member who leaked the code to me. Perhaps you’ve grown frustrated with staff’s increasing opaqueness over the past several years. We did too.
We decided to develop a simple tool to collect weekly update data from the website to see if we could spot any trends in things like request resolution times, character applications, and of course, player numbers. We worked from 2014 forward since 2013 has a huge gap in its dataset, presumably due to a database error. Besides, six years of recent data should be plenty to form some trends and opinions with.
This is going to be a multi-part series where we go over some charts and tables and make some theories on what’s happened so far and where things can go in the future. I’m not a data scientist, but I am familiar enough with analyses like this from my business experience. I like to think I’m familiar enough with this stuff to know what a good trendline looks like. This series of posts is not a criticism of Armageddon, but rather, a discussion on what has happened to it and what can be redeemed. On that note, I’m not going to discuss the “symptoms” of whatever we might find in the original post, i.e. talk what game-related or community-related matters can explain the information, until the end; I’m here to just present information.
This is going to be a long post. If you’re not interested, that’s okay. I’m mostly doing this and posting this as a favor to the aforementioned staff member, and as a resource for the broader community. That’s not to say I’m uninterested, but it’s just to help explain motives. There will be a link containing the data we collected at the end of the entire series, and I’ll probably post one part every week or so.
Definitions
Before we go further, we need to understand what metrics like this actually mean:
There has been a lot of confusion in the past about how those numbers are generated and calculated. The leaked code tells us that login information is only saved into the game’s database when you log into a character. This is likely where the “unique logins” number comes from - the site simply appears to query the database for the number of unique accounts that have logged in that week and uses JavaScript to put it on the page. The “new accounts created” number is a bit more cryptic, it seems. The code is hard to read through, but there doesn’t actually seem to be any code to keep track of when a new account is created. We’re left to assume that the “new accounts created” number is actually the number of accounts who log in for the first time with their first character. But it’s not visible in the code, and that condition is close enough to “new accounts created” that it should not actually be a problem.
Data was collected by week and by year. The weekly updates URL is in this format:
armageddon.org/updates/index.php?week=X&year=Y
Where X is the week number (usually 1 - 52) and Y is the year (2014 - 2019). So if you see “34/2016” on a chart, that means that data point is associated with the 34th week of 2016, and if you want the specific dates you can fill in the variables in the URL and load it on a browser to see that it was August 22 to August 29, 2016. Note that if the week you’re looking for is a single digit, you will need to append a 0 to the beginning (if you want to look at the 5th week of 2018, you’d type “05” for the week).
If you see extra space on the graph where a trendline is leading, that’s forecasting what Armageddon’s next 52 weeks might look like - or in other words, what 2020 could look like.
The equation and R-squared values on the chart correspond to the trendline. The equation is literally the equation of the trendline itself, and the R-squared value is basically a percentage of how well the trendline matches up with the data we have. The closer R-squared is to 1 (100%), the closer the trendline conforms to our data, though that does not necessarily mean it is more accurate: for example, it’s entirely possible to have an R-squared value of 1 for a trendline that shows a trend leading to negative logins in 2020, which is obviously impossible.
Unique Logins (2014 - 2019)
This graph shows all of the weekly logins from the beginning of 2014 to the end of 2019. The solid line shows the actual data and the dotted line is a trendline. The best possible trendline I thought was reliable in this case was a cubic function, and you can see that it’s gradually leveling off, like the data appears to be, just below the 190 mark. I suspect that it will dip a little bit before flatlining - making more significant predictions than that beyond 2020 is impractical, but all cubic functions like this will eventually dip again, so it’s obviously not a perfect model for the long run. It seems to be good enough for the data we have though. Regardless of the exact accuracy of the trendline, there is a clear downward trend in logins that doesn’t seem to be ending anytime soon, although the decrease will be less pronounced than in previous years. Having been in the MUD world, though, I know a lot of people decide whether to stay logged in based on the number of people logged in when they log in themselves. So I suspect that when the player count reaches a certain point, the decline will actually accelerate as more people find there are fewer people to play with.
It’s worth taking at the yearly averages to see a more broad trend:
We see that, if we average the logins for all weeks of any given year, there is a progressive downward trend that is likely to result in around 185 players logging in weekly on average in 2020.
New Accounts Created
This chart shows the new accounts created from 2014 to 2019. It’s really hard to build a reliable trendline for this 6-year dataset and I would argue that it’s not very accurate, but the data itself shows a pretty clear declining trend in interest in even trying the game out over time. But let’s take a look at the yearly averages:
From 2014-2016, the average seems pretty stable, but tanks after that. I can’t really offer an explanation as to why, but it’s important to note that Armageddon was delisted from Top Mud Sites for vote-cheating in 2018, so that doesn’t quite explain the drop. I tried to be as generous as possible with the trendline I used, but it’s very possible there will be weeks in 2020 where no new players log in. The current trend shows an average of about 5 accounts a week per year in 2020.
What really matters in the long-term, though, are the total amount of new accounts over the course of six years. So I graphed that too:
There were 12,006 (!) new signups over the course of 6 years, but it follows a parabolic trend that isn’t surprising considering the previous data that shows things leveling off. For whatever reason, the game is running out of new people to recruit.
The question ultimately becomes: out of these 12,006 new accounts, how many players is the game actually keeping?
Retained Players
There’s effectively no way to see exactly how many non-new players are logging in every week. But we can estimate it by subtracting the number of new accounts on a given week from the number of weekly logins. This is ultimately what we see:
To be honest, I’m not a huge supporter of the trendline’s accuracy. I think the data is way too scattered to draw any long-term trend from it. But given that the new player count is apparently leveling off per the previous charts, it makes sense that the veteran player count should eventually level off - particularly in the 170ish range, if this graph is to be believed.
There’s nothing much else to say about this particular statistic. Overall logins are down, and overall new signups are down, so it makes sense that the people logging in regularly is leveling off for now. This number won’t really drop until the new signup count is zero or close to zero for a long length of time. This is probably the most interesting stat to check back on in 2021, though, because of the yearly averages:
There is a clear upward trend that seems to defy the unique logins trend discussed earlier, although I don’t think a simple parabola is going to show us the correct trend for a year out. The real trend is probably somewhere between this parabola and the last chart’s logarithmic curve. Still, I left it in to give some hope to people who think Armageddon can rise from the hole it’s in now, player counts-wise, and because other trendlines are more bleak considering the upward move the past couple of years, or highly unrealistic (e.g., negative players in 2020).
More realistically, I suspect that it will start to level off once the new player count starts to regularly hover in the single digits, and then slowly drop again. It’s also quite possible that the upward trend is due to players returning to the game and using their old accounts to do so, in which case the data can be expected to level off for similar reasons. We’ve already seen that there were 12,006 new accounts created in 6 years, yet in most of that time, established player logins dropped, and is only starting to increase now. Overall, the game has negative player retention, apparently having lost about 10% of its players in the past six years, If we just look at the 4088 new accounts that signed up from 2017 to 2019, where the upward established player trend exists, then the game retained about eight of those players on average, or about 0.195% of its new players. In case it’s not obvious, I should stress that this is extremely negligible, especially considering the possibility that old players simply returned to the game in this same time period.
Conclusions
We can see clearly that Armageddon is not doing well, as far as player numbers go. But why? I said earlier that I’m not interested in discussing game-related symptoms such as game changes, or community-related symptoms such as aggressive staff or players, until I’ve presented all of the data. So now that we’re here, we can get into a few of the common explanations for why Armageddon’s player data looks to be in decline and try to see if those explanations match up with the data.
“It’s the holidays! No one plays on the holidays”: Half-True
If we detrend all of the unique login data and overlap each year’s data with one another, and measure week-to-week changes in logins on the vertical axis, we can try to find common yearly trends that can be explained by the time of year.
The result is quite messy, but if you look closely enough you can see some common dips in the player numbers. For example, around week 36-37 of each year the numbers always decrease. It’s almost as clear around week 15 and week 48. Week 36 - 37 is around and after Labor Day; Week 48 is around Thanksgiving; Week 15 is generally around Easter. Some years show similar dips around the end of December while others don’t. It makes sense that the player numbers would dip around these times of year every year, but it doesn’t explain an overall six-year decline in unique logins.
So while it’s true that fewer players play during the holidays, and the numbers do eventually rebound slightly, it’s just not enough of a rebound to counteract the larger declining trend in player numbers.
“New players are down because TMS delisted Arm because of cheating goblins”: Probably False
Nathvaan announced that Top Mud Sites delisted Armageddon on February 1, 2018, citing vote inflation practices that are considered cheating by the site. That corresponds to the 5th week of 2018. If we look at our charts we can see that there is no particular change in the data set around that time. It takes about 10 weeks for the new account data to suffer a sudden drop, and unique login data seems to spike before steadily dropping over the rest of the year. If there really was a troll trying to ruin Armageddon’s exposure, they failed in this particular task, because new players were able to find the game undeterred for about two and a half months afterward. And if it was just an overzealous player cheating to help the game, their mistake didn’t really do much to break anything - and quite possibly helped for the prior days and/or weeks that they were able to cheat and get away with it, depending on how long the cheater(s) was/were working.
“New players are down because of bad/fake reviews”: Probably False
My friend and I went through all posts related to Armageddon MUD on Reddit as well as reviews on Mudconnect, briefly read through them, and assigned a “Positive” or “Negative” rating to the weeks in which they appeared on those sites based on whether they seemed to be positive or negative in nature. If the review was mixed, we rated the week as “Mixed” and if there were multiple posts/reviews that were positive and negative on the same week, we also rated the week “Mixed”. Obviously, this rating system can be pretty subjective, so we tried to look for clear-cut phrases like “Come join us” or “Don’t play this game” to determine a rating. If we couldn’t find a call to action like that, we judged the post by its tone and content: as a contrived example, if a post said the player was harassed while playing the game, we rated it “Negative”. For posts that didn’t directly promote Armageddon, but gave its theme and roleplaying setting some exposure, such as ShaLeah’s promotional posts on Reddit around 2018, we gave those a Positive rating since they’re intended to attract positive attention to the game. We overlaid this over the chart we have for new account creation per-week.
There are plenty of dips in new account creation after positive reviews, and increases in new account creation after negative reviews, and varying trends after mixed reviews, that I think it’s safe to say the kind of review doesn’t have much of a reliable effect on new player count at all. Big clusters of reviews and posts result in spikes in attention but those spikes have progressively shrunk. This is probably unrelated to the reviews themselves, but rather the entire pool of MUD players without a MUD home shrinking - in other words, most of the people who are interested in MUDs and would be interested in Armageddon have already checked it out. Exposure of any kind - positive or negative - in large amounts used to be good for Armageddon a couple of years ago, but nowadays it barely has any effect at all. This is unfortunate considering how many players volunteer to post positive things about Armageddon.
“There are plenty of new faces around, therefore player retention is good”: Not necessarily true
I don’t really have a chart to support this directly, but I will direct you back to my explanation on the retained player trend and add that just because people are joining the game doesn’t necessarily mean people aren’t leaving the game at the same time. Think of a bus with 100 passengers. If, every stop it makes, 10 passengers get off the bus and 10 more get on, there will be a decent amount of new passengers on the bus after every bus stop, but there will also be people who have left the bus after riding a long time. After enough stops, the composition of the entire group of passengers will be completely different. I think Armageddon’s a lot like that bus right now, except more people are getting on than off. But just a little bit more, and only within the past couple of years. New player retention might be good, but if veteran player retention is bad, then overall retention evens out to being negligible. The solution to Armageddon’s player retention problem is not just to address whether newbies are staying aboard, but whether veterans are staying aboard too, and how to cater to both groups.
“X Major Change to the game improved/depressed numbers”: It’s complicated
There’s a lot of changes we can point to in the past six years - like Tuluk’s closure, the class system change, and the karma system change - that we can align with the various charts and see if they had any measurable effect. And while sometimes these changes resulted in a short spike in login numbers (for a few weeks or so), the numbers continued on their overall downward trend. Individually, they have made no noticeable impact at all on new player signups. It is possible that a combination of these changes contributed to a sort of “last straw” situation for some players, but there’s no way to confirm it with data.
Conclusions that can’t be measured
Conclusions that can’t be measured are probably thrown around the most often, particularly by staff in staff-to-player communications that have come to light over the past few years; it happens to be convenient to people who peddle these theories that there is no way to measure these conclusions against hard data. Theories like "shadowboarders are harming the game" are not measurable. That said, if they think of a way to measure such data, they can do so and are free to compare their findings to the charts I have (and, once this thread series is done, the data tables).
The End of Part 1
So we’ve explored some pretty common explanations for the falling numbers. What actually explains it? That’s a good question, and that’s a question I’ll turn on anyone here who wants to answer. I will say if no one can find a correlation between some measurable statistics and these statistics on player numbers, then it’s probably something that can’t be easily measured - like a general satisfaction level with the game, which would require Armageddon staff to survey its current and past players and report on the results accurately. And that introduces a ton of selection bias - who’s more likely to answer such a survey, a current and longtime player or someone who’s left the game a long time ago?
If you’re thinking about replying but don’t know what to say, here are some prompts. Feel free to use them if you’d like, or address another point. I will write some opinions on these questions, just probably later:
And of course, if you have any questions on the presented charts, feel free to ask.
Special thanks to the 2016 code leaker (who still wishes to remain anonymous, for now, at least) for coming up with the idea for this and a lot of the technical resources that would be needed in order to achieve it. And a big thanks to none other than Armageddon’s website coders for keeping these numbers up on the website in the first place. Even though the game appears to be in decline, they are at least transparent about that for now, even if they choose not to comment on it or do anything of note about it.
After a week or so I’ll probably come up with a thread on a different aspect of the data. Are people more interested in request resolution or character apps?
If you care or have cared at all about Armageddon in the past, you’ve probably thought about the game’s current state and what things might look like in the future. Well, I did, at least. And so did my friend, the former staff member who leaked the code to me. Perhaps you’ve grown frustrated with staff’s increasing opaqueness over the past several years. We did too.
We decided to develop a simple tool to collect weekly update data from the website to see if we could spot any trends in things like request resolution times, character applications, and of course, player numbers. We worked from 2014 forward since 2013 has a huge gap in its dataset, presumably due to a database error. Besides, six years of recent data should be plenty to form some trends and opinions with.
This is going to be a multi-part series where we go over some charts and tables and make some theories on what’s happened so far and where things can go in the future. I’m not a data scientist, but I am familiar enough with analyses like this from my business experience. I like to think I’m familiar enough with this stuff to know what a good trendline looks like. This series of posts is not a criticism of Armageddon, but rather, a discussion on what has happened to it and what can be redeemed. On that note, I’m not going to discuss the “symptoms” of whatever we might find in the original post, i.e. talk what game-related or community-related matters can explain the information, until the end; I’m here to just present information.
This is going to be a long post. If you’re not interested, that’s okay. I’m mostly doing this and posting this as a favor to the aforementioned staff member, and as a resource for the broader community. That’s not to say I’m uninterested, but it’s just to help explain motives. There will be a link containing the data we collected at the end of the entire series, and I’ll probably post one part every week or so.
Definitions
Before we go further, we need to understand what metrics like this actually mean:
There has been a lot of confusion in the past about how those numbers are generated and calculated. The leaked code tells us that login information is only saved into the game’s database when you log into a character. This is likely where the “unique logins” number comes from - the site simply appears to query the database for the number of unique accounts that have logged in that week and uses JavaScript to put it on the page. The “new accounts created” number is a bit more cryptic, it seems. The code is hard to read through, but there doesn’t actually seem to be any code to keep track of when a new account is created. We’re left to assume that the “new accounts created” number is actually the number of accounts who log in for the first time with their first character. But it’s not visible in the code, and that condition is close enough to “new accounts created” that it should not actually be a problem.
Data was collected by week and by year. The weekly updates URL is in this format:
armageddon.org/updates/index.php?week=X&year=Y
Where X is the week number (usually 1 - 52) and Y is the year (2014 - 2019). So if you see “34/2016” on a chart, that means that data point is associated with the 34th week of 2016, and if you want the specific dates you can fill in the variables in the URL and load it on a browser to see that it was August 22 to August 29, 2016. Note that if the week you’re looking for is a single digit, you will need to append a 0 to the beginning (if you want to look at the 5th week of 2018, you’d type “05” for the week).
If you see extra space on the graph where a trendline is leading, that’s forecasting what Armageddon’s next 52 weeks might look like - or in other words, what 2020 could look like.
The equation and R-squared values on the chart correspond to the trendline. The equation is literally the equation of the trendline itself, and the R-squared value is basically a percentage of how well the trendline matches up with the data we have. The closer R-squared is to 1 (100%), the closer the trendline conforms to our data, though that does not necessarily mean it is more accurate: for example, it’s entirely possible to have an R-squared value of 1 for a trendline that shows a trend leading to negative logins in 2020, which is obviously impossible.
Unique Logins (2014 - 2019)
This graph shows all of the weekly logins from the beginning of 2014 to the end of 2019. The solid line shows the actual data and the dotted line is a trendline. The best possible trendline I thought was reliable in this case was a cubic function, and you can see that it’s gradually leveling off, like the data appears to be, just below the 190 mark. I suspect that it will dip a little bit before flatlining - making more significant predictions than that beyond 2020 is impractical, but all cubic functions like this will eventually dip again, so it’s obviously not a perfect model for the long run. It seems to be good enough for the data we have though. Regardless of the exact accuracy of the trendline, there is a clear downward trend in logins that doesn’t seem to be ending anytime soon, although the decrease will be less pronounced than in previous years. Having been in the MUD world, though, I know a lot of people decide whether to stay logged in based on the number of people logged in when they log in themselves. So I suspect that when the player count reaches a certain point, the decline will actually accelerate as more people find there are fewer people to play with.
It’s worth taking at the yearly averages to see a more broad trend:
We see that, if we average the logins for all weeks of any given year, there is a progressive downward trend that is likely to result in around 185 players logging in weekly on average in 2020.
New Accounts Created
This chart shows the new accounts created from 2014 to 2019. It’s really hard to build a reliable trendline for this 6-year dataset and I would argue that it’s not very accurate, but the data itself shows a pretty clear declining trend in interest in even trying the game out over time. But let’s take a look at the yearly averages:
From 2014-2016, the average seems pretty stable, but tanks after that. I can’t really offer an explanation as to why, but it’s important to note that Armageddon was delisted from Top Mud Sites for vote-cheating in 2018, so that doesn’t quite explain the drop. I tried to be as generous as possible with the trendline I used, but it’s very possible there will be weeks in 2020 where no new players log in. The current trend shows an average of about 5 accounts a week per year in 2020.
What really matters in the long-term, though, are the total amount of new accounts over the course of six years. So I graphed that too:
There were 12,006 (!) new signups over the course of 6 years, but it follows a parabolic trend that isn’t surprising considering the previous data that shows things leveling off. For whatever reason, the game is running out of new people to recruit.
The question ultimately becomes: out of these 12,006 new accounts, how many players is the game actually keeping?
Retained Players
There’s effectively no way to see exactly how many non-new players are logging in every week. But we can estimate it by subtracting the number of new accounts on a given week from the number of weekly logins. This is ultimately what we see:
To be honest, I’m not a huge supporter of the trendline’s accuracy. I think the data is way too scattered to draw any long-term trend from it. But given that the new player count is apparently leveling off per the previous charts, it makes sense that the veteran player count should eventually level off - particularly in the 170ish range, if this graph is to be believed.
There’s nothing much else to say about this particular statistic. Overall logins are down, and overall new signups are down, so it makes sense that the people logging in regularly is leveling off for now. This number won’t really drop until the new signup count is zero or close to zero for a long length of time. This is probably the most interesting stat to check back on in 2021, though, because of the yearly averages:
There is a clear upward trend that seems to defy the unique logins trend discussed earlier, although I don’t think a simple parabola is going to show us the correct trend for a year out. The real trend is probably somewhere between this parabola and the last chart’s logarithmic curve. Still, I left it in to give some hope to people who think Armageddon can rise from the hole it’s in now, player counts-wise, and because other trendlines are more bleak considering the upward move the past couple of years, or highly unrealistic (e.g., negative players in 2020).
More realistically, I suspect that it will start to level off once the new player count starts to regularly hover in the single digits, and then slowly drop again. It’s also quite possible that the upward trend is due to players returning to the game and using their old accounts to do so, in which case the data can be expected to level off for similar reasons. We’ve already seen that there were 12,006 new accounts created in 6 years, yet in most of that time, established player logins dropped, and is only starting to increase now. Overall, the game has negative player retention, apparently having lost about 10% of its players in the past six years, If we just look at the 4088 new accounts that signed up from 2017 to 2019, where the upward established player trend exists, then the game retained about eight of those players on average, or about 0.195% of its new players. In case it’s not obvious, I should stress that this is extremely negligible, especially considering the possibility that old players simply returned to the game in this same time period.
Conclusions
We can see clearly that Armageddon is not doing well, as far as player numbers go. But why? I said earlier that I’m not interested in discussing game-related symptoms such as game changes, or community-related symptoms such as aggressive staff or players, until I’ve presented all of the data. So now that we’re here, we can get into a few of the common explanations for why Armageddon’s player data looks to be in decline and try to see if those explanations match up with the data.
“It’s the holidays! No one plays on the holidays”: Half-True
If we detrend all of the unique login data and overlap each year’s data with one another, and measure week-to-week changes in logins on the vertical axis, we can try to find common yearly trends that can be explained by the time of year.
The result is quite messy, but if you look closely enough you can see some common dips in the player numbers. For example, around week 36-37 of each year the numbers always decrease. It’s almost as clear around week 15 and week 48. Week 36 - 37 is around and after Labor Day; Week 48 is around Thanksgiving; Week 15 is generally around Easter. Some years show similar dips around the end of December while others don’t. It makes sense that the player numbers would dip around these times of year every year, but it doesn’t explain an overall six-year decline in unique logins.
So while it’s true that fewer players play during the holidays, and the numbers do eventually rebound slightly, it’s just not enough of a rebound to counteract the larger declining trend in player numbers.
“New players are down because TMS delisted Arm because of cheating goblins”: Probably False
Nathvaan announced that Top Mud Sites delisted Armageddon on February 1, 2018, citing vote inflation practices that are considered cheating by the site. That corresponds to the 5th week of 2018. If we look at our charts we can see that there is no particular change in the data set around that time. It takes about 10 weeks for the new account data to suffer a sudden drop, and unique login data seems to spike before steadily dropping over the rest of the year. If there really was a troll trying to ruin Armageddon’s exposure, they failed in this particular task, because new players were able to find the game undeterred for about two and a half months afterward. And if it was just an overzealous player cheating to help the game, their mistake didn’t really do much to break anything - and quite possibly helped for the prior days and/or weeks that they were able to cheat and get away with it, depending on how long the cheater(s) was/were working.
“New players are down because of bad/fake reviews”: Probably False
My friend and I went through all posts related to Armageddon MUD on Reddit as well as reviews on Mudconnect, briefly read through them, and assigned a “Positive” or “Negative” rating to the weeks in which they appeared on those sites based on whether they seemed to be positive or negative in nature. If the review was mixed, we rated the week as “Mixed” and if there were multiple posts/reviews that were positive and negative on the same week, we also rated the week “Mixed”. Obviously, this rating system can be pretty subjective, so we tried to look for clear-cut phrases like “Come join us” or “Don’t play this game” to determine a rating. If we couldn’t find a call to action like that, we judged the post by its tone and content: as a contrived example, if a post said the player was harassed while playing the game, we rated it “Negative”. For posts that didn’t directly promote Armageddon, but gave its theme and roleplaying setting some exposure, such as ShaLeah’s promotional posts on Reddit around 2018, we gave those a Positive rating since they’re intended to attract positive attention to the game. We overlaid this over the chart we have for new account creation per-week.
There are plenty of dips in new account creation after positive reviews, and increases in new account creation after negative reviews, and varying trends after mixed reviews, that I think it’s safe to say the kind of review doesn’t have much of a reliable effect on new player count at all. Big clusters of reviews and posts result in spikes in attention but those spikes have progressively shrunk. This is probably unrelated to the reviews themselves, but rather the entire pool of MUD players without a MUD home shrinking - in other words, most of the people who are interested in MUDs and would be interested in Armageddon have already checked it out. Exposure of any kind - positive or negative - in large amounts used to be good for Armageddon a couple of years ago, but nowadays it barely has any effect at all. This is unfortunate considering how many players volunteer to post positive things about Armageddon.
“There are plenty of new faces around, therefore player retention is good”: Not necessarily true
I don’t really have a chart to support this directly, but I will direct you back to my explanation on the retained player trend and add that just because people are joining the game doesn’t necessarily mean people aren’t leaving the game at the same time. Think of a bus with 100 passengers. If, every stop it makes, 10 passengers get off the bus and 10 more get on, there will be a decent amount of new passengers on the bus after every bus stop, but there will also be people who have left the bus after riding a long time. After enough stops, the composition of the entire group of passengers will be completely different. I think Armageddon’s a lot like that bus right now, except more people are getting on than off. But just a little bit more, and only within the past couple of years. New player retention might be good, but if veteran player retention is bad, then overall retention evens out to being negligible. The solution to Armageddon’s player retention problem is not just to address whether newbies are staying aboard, but whether veterans are staying aboard too, and how to cater to both groups.
“X Major Change to the game improved/depressed numbers”: It’s complicated
There’s a lot of changes we can point to in the past six years - like Tuluk’s closure, the class system change, and the karma system change - that we can align with the various charts and see if they had any measurable effect. And while sometimes these changes resulted in a short spike in login numbers (for a few weeks or so), the numbers continued on their overall downward trend. Individually, they have made no noticeable impact at all on new player signups. It is possible that a combination of these changes contributed to a sort of “last straw” situation for some players, but there’s no way to confirm it with data.
Conclusions that can’t be measured
Conclusions that can’t be measured are probably thrown around the most often, particularly by staff in staff-to-player communications that have come to light over the past few years; it happens to be convenient to people who peddle these theories that there is no way to measure these conclusions against hard data. Theories like "shadowboarders are harming the game" are not measurable. That said, if they think of a way to measure such data, they can do so and are free to compare their findings to the charts I have (and, once this thread series is done, the data tables).
The End of Part 1
So we’ve explored some pretty common explanations for the falling numbers. What actually explains it? That’s a good question, and that’s a question I’ll turn on anyone here who wants to answer. I will say if no one can find a correlation between some measurable statistics and these statistics on player numbers, then it’s probably something that can’t be easily measured - like a general satisfaction level with the game, which would require Armageddon staff to survey its current and past players and report on the results accurately. And that introduces a ton of selection bias - who’s more likely to answer such a survey, a current and longtime player or someone who’s left the game a long time ago?
If you’re thinking about replying but don’t know what to say, here are some prompts. Feel free to use them if you’d like, or address another point. I will write some opinions on these questions, just probably later:
- What can be done to attract people who are not necessarily interested in/don’t know about MUDs to the game? What can be done to pull people in from the MUD community who may or may not already have a preferred MUD?
- Is there something about Armageddon that makes it difficult to recommend to people via word-of-mouth (reviews, promotions, etc.)?
- What do you think will happen in 2020 with these numbers? Will they follow the trends I established or do you think they’ll do something else?
- What should staff do to improve the data they post on the site? What should they actually do with that data, if anything?
And of course, if you have any questions on the presented charts, feel free to ask.
Special thanks to the 2016 code leaker (who still wishes to remain anonymous, for now, at least) for coming up with the idea for this and a lot of the technical resources that would be needed in order to achieve it. And a big thanks to none other than Armageddon’s website coders for keeping these numbers up on the website in the first place. Even though the game appears to be in decline, they are at least transparent about that for now, even if they choose not to comment on it or do anything of note about it.
After a week or so I’ll probably come up with a thread on a different aspect of the data. Are people more interested in request resolution or character apps?
