Cogg demographics

Have a new general feature to suggest, or think one should be tweaked? Share your ideas here.
User avatar
nobody
Posts: 501
Joined: Wed Jun 17, 2020 7:43 am
Contact:

Cogg demographics

Post by nobody »

Please, please, feel free to nope this. That caveat stated, I'd love to be able to look at the character demographics to hopefully draw some conclusions from looking at the data in a high level way. To that end, I was wondering if I might request a data source similar in style to the weekly changelog (http://cogg.contrarium.net/changelog/weekly.html), but with some very high level character data (and maybe dumped to http://cogg.contrarium.net/demographics/weekly.html).
Below is an example very made up dataset (only has Active (A), not Shelved (S) or Retired (R) characters)

Code: Select all

class,status,count,level,xp
bard,A,51,853,6912399
ranger,A,202,7198,82860524
treasure_hunter,A,53,425,2659444
rogue,A,53,1678,17473034
arcanist,A,27,56,229228
primalist,A,20,35,131288
physicker,A,29,37,112654
warlock,A,34,362,2369426
berserker,A,90,576,3384442
dreadnought,A,15,112,717476
duelist,A,89,538,3147521
guardian,A,123,4633,54724811
marauder,A,124,941,6008222
nightblade,A,22,237,1608361
I deliberately limited it to very high level data (count, sum of levels, sum of xp) for each class because our player base is small enough that even a comma-separated list of levels or xp would probably identify specific individual characters (which seems not cool). For a similar reason, rolling/daily data seems more invasive than weekly data. Example insights from the (very made up) data set above include but are not limited to
  • ranger is the most popular class by count and this is also true by level and XP
  • rogues and treasure hunters look tied by count but when level and XP are factored in, rogues are vastly more popular. Same deal for guardian vs marauder
  • the average level for arcanists, primalists, and physickers are all under 3, suggesting people try them out, but hit a wall pretty early
The above type of insights can be gained from a single dataset, but I think where this would really shine is with weekly data updates. Comparing two data sets over time would show how much XP was gained by various classes (IE, what classes are active in that time period) and how counts shifted.

All that being said, there are many things that such data won't come close to touching, like the quantity and quality of RP happening in game. I still think it has value though, hence the request. Thanks for reading.
User avatar
Rias
DEV
Posts: 2024
Joined: Sun Sep 03, 2017 4:06 pm
Location: Wandering Temicotli

Re: Cogg demographics

Post by Rias »

I've been doing a lot with APIs at work lately and naturally gravitated to this request as a result.

https://cogg.contrarium.net/demographics/

I generate it on manual command at the moment, but a weekly Monday update to match the weekly changelog would work well, I think. There's a link there to get the raw JSON data if anyone wants to do whatever with it. Because data is fun!

Feel free to share any comments or insights based on the data. Like the fact that there are a huge number of fence-sitters who haven't chosen a guild yet, and that physickers are apparently quite dedicated.

P.S. I'll probably add a toggle for active (not shelved/retired) vs all characters, but I didn't feel comfortable with that just yet while the population is still small and some people could be rather easily singled out to essentially snoop their data.
<Rias> PUT ON PANTS
<Fellborn> NO
User avatar
nobody
Posts: 501
Joined: Wed Jun 17, 2020 7:43 am
Contact:

Re: Cogg demographics

Post by nobody »

So this was more data than I had dreamed of and it's been great fun looking at it. I decided to tackle the low-hanging fruit first, and that is skills. I've included some excerpt tables, but posted them at the bottom of the post to try to make things a little easier on our screen reader users (the tables aren't friendly for that I think).

Most popular by ranks (Table 1): The top six skills by rank are stealth, medical, melee combat, dodge, armor use, and perception. This probably shouldn't come as a surprise, physickers have been outed as the most popular class by far for a while, and everything else is combat or combat adjacent.

Most popular by points (Table 2): The top six skills by points are medical, stealth, arcana, melee combat, dodge, and bushcraft. The comparison is very interesting because this category shows depth of investment, rather than breadth. Medical was expected, stealth, melee combat, and dodge were all expected, but arcana and bushcraft are both delightful surprises to me, especially since arcana is not very fleshed out yet. Clearly people are excited about arcana.

Ordered by the ratio of points to ranks (Tables 3 and 4): So here is where the fun starts in my opinion. At the top of the table we have lapidary, medical, construction... wait, lapidary? Hang on there, we have to just address that first: lapidary doesn't do anything yet (I think). I very briefly tried to find where I could train it and came up blank, so this is maybe test characters of our lovely dev. Hi Rias! Alright, that brief outing aside, at the top of the table we have medical, construction, metalworking, animal husbandry, riding, and druidry while at the bottom of the table we have two distinct groups: trading, music, and linguistics (not fully implemented), and metallurgy, shield use, leatherworking, and pottery. Medical makes sense, physickers are the most common class specialization by count, level, and XP, that means that people make physickers, and they level up their physickers, and they dump at least some of those points into advanced ranks in medical despite there not being a lot of incentive to do so. Construction, metalworking, and animal husbandry all make sense, those are skills that have mechanically coded benefits for heavier skill investment (lots of recipes, desirable recipes that have high skill requirements, and for animal husbandry that juicy, juicy land allotment (plus access to different animals I think?)). Druidry is a delightful surprise, but probably shouldn't have been. It's interesting, has some neat lore and questing to do, and does offer perks for very high skill (longer duration and added effects for some abilities). Riding is a surprise, and I suspect a fluke from a small number of high-level players that have been planning on riding in their builds, but I could easily be wrong there. We see that stealth isn't near the top of the table any more, which means most of those stealth ranks are spread out across a LOT of characters - that makes sense, it's a fun skill to have even if you only have a little.

The skills at the bottom of the table (that are implemented) also make some sense. Metallurgy and shield use are secondary or helper skills and insofar as I know, metallurgy doesn't reward deep investment. Pottery and leatherworking are interesting cases. On the up side, pottery has been around for quite a while, it's very inexpensive to get started in (you only need a shovel and a bucket and for mass production maybe your own table or handcart) and it makes useful things. On the down side, it doesn't have a lot of recipes and only one requires advanced ranks to produce and that item can be bought directly from stores, and nothing it produces can be worn or used in combat (aside from sling bullets that can be purchased directly or crafted with two other skills). It has not seen a lot of investment though, and the difference between points and ranks is only 75, which suggest that no character in the game can produce the one high-skill recipe. Leatherworking on the other hand was difficult to raise for a long time, but that got modified at the same time as tailoring. It doesn't lack for recipes, or steep skill requirements on those recipes. It is currently the only way to produce armor, a highly sought after good. Given all that, why doesn't it see the same kind of investment as say metal working or tailoring (which got improved at the same time, and for reference has a point-to-rank ratio of 1.53 compared to leatherworking's 1.08)? I suspect it might be the depth of skill investment required to gather hides compared to wool or ore. To collect hides, one needs stealth, tracking, perception, ranged (or melee) combat, and skinning, while collecting ore only requires mining and wool only requires animal husbandry. I suspect leatherworking will eventually see more investment, especially as players have been making hides more available at market, but I hypothesize people have held off either because of lack of available materials in the market or the relatively higher skill cost of controlling the production process from start to finish. That's just my guess though, and it's been fun looking at the data and thinking about it.

Ratio of ranks to goals (no table): I looked at this briefly and gave up on it fairly quickly after seeing weaving at 1.32 - of the planned 2100 ranks, 2785 have been attained. Hmm. Yeah that could be because it's been around for a while and not everyone uses the goals command to plan their stuff. Not much use in looking at the goals in ratios then. I basically stopped looking at goals data there, but looking at the skills ordered by goals is still interesting (combat and exploration at the top, some lesser used or secondary crafts at the bottom).

Other measures (also no table): I've been looking at the table and poking around with the numbers to see if anything stands out. One of the things I looked at was a sort of advanced-rank-point ratio, where I take the difference between points and ranks giving us the maximum number of ranks between 101-200. Pottery is the easy example, 833 points 758 ranks, 75 advanced ranks, max. Those 75 ranks all must fall between 101-200 because we'd need at least 101 advanced ranks for that to not be true, and so they cost 2 points each. Thus our ratio is 2 x (points - ranks) / points, or 0.18 for pottery. That is, 18% of invested skill points in pottery are advanced ranks, while 72% are from introductory (1-100) ranks. The ordering is identical to the points to ranks ratio, but this feels like it's telling us something more interesting because it has an actual interpretation. It doesn't hold well though, for example medical, construction, and metal working are all so heavily invested that they break this metric. Medical's ratio is 1.07... so 107% of invested points are in advanced ranks... no, no, obviously that's not true. It's easy enough to understand why it breaks, ranks above 200 cost 3 instead of 2 so the math has to change, but with most skills it's pretty impossible to determine how the points break out (and that's probably good, that potter (or potters) with pottery above 100 probably feels pretty seen. Sorry.). On the other hand, all you people that aspire to the best at something, look at the bottom of the table, there's lots of room for people to climb to the top from down there.

Code: Select all

Table 1 (ordered by ranks)
    skills      ranks   goals   points   points/ranks   ranks/goals 
-------------- ------- ------- -------- -------------- -------------
   Stealth      5653    17620    7471       1.322         0.3208    

   Medical      5628    8051    12085       2.147          0.699    

 Melee Combat   5379    21375    7062       1.313         0.2516    

    Dodge       5222    16995    6774       1.297         0.3073    

  Armor Use     4445    8810     5900       1.327         0.5045    

  Perception    4402    19420    5832       1.325         0.2267    
--------------------------------------------------------------------

Code: Select all

Table 2 (ordered by points)
    skills      ranks   goals   points   points/ranks   ranks/goals 
-------------- ------- ------- -------- -------------- -------------
   Medical      5628    8051    12085       2.147          0.699    

   Stealth      5653    17620    7471       1.322         0.3208    

    Arcana      4336    11683    7459        1.72         0.3711    

 Melee Combat   5379    21375    7062       1.313         0.2516    

    Dodge       5222    16995    6774       1.297         0.3073    

  Bushcraft     4089    5750     6216        1.52         0.7111    
--------------------------------------------------------------------

Code: Select all

Table 3 (ordered by points/ranks)
      skills        ranks   goals   points   points/ranks   ranks/goals 
------------------ ------- ------- -------- -------------- -------------
     Lapidary        494     450     1470       2.976          1.098    

     Medical        5628    8051    12085       2.147          0.699    

   Construction     1841    2500     3763       2.044         0.7364    

   Metalworking     2973    3500     6051       2.035         0.8494    

 Animal Husbandry   2942    4200     5817       1.977         0.7005    

      Riding         740    5400     1440       1.946          0.137    
------------------------------------------------------------------------

Code: Select all

Table 4 (ordered by points / ranks, the bottom of the table)
     skills       ranks   goals   points   points/ranks   ranks/goals 
---------------- ------- ------- -------- -------------- -------------
    Pottery        758     800     833        1.099         0.9475    

 Leatherworking    424    2500     458         1.08         0.1696    

   Shield Use      905    4100     975        1.077         0.2207    

   Metallurgy      748    2100     798        1.067         0.3562    

    Trading        145    3300     145          1           0.04394   

     Music         30     3200      30          1          0.009375   

  Linguistics      703    10000    703          1           0.0703    
----------------------------------------------------------------------
Edit to add: Fun fact! There are 43,003 unspent skill points for the collective player base at the time of the data posting - about 108 per character on average.
User avatar
Lexx416
Posts: 563
Joined: Wed Sep 06, 2017 6:14 pm

Re: Cogg demographics

Post by Lexx416 »

Having done the pottery thing on my character, I believe I ended up losing interest because of the lack of overall use for pottery. Clay bullets are useful, and I've made plenty. But I've also got like... 50 or 60 of those iron skulls that I use as bullets instead. So there's just not a lot of reason for me to go with clay over iron. I also think that the gathering of clay and water to use in the clay works can be tedious, and that put me off of it as well.

I am kind of surprised that there isn't more leather working. Maybe the tanning process is putting people off. I stopped playing for a while, but my character as long as I've played them has offered hides and pelts for sale.
"You hear the Woses, the Wild Men of the Woods... Remnants of an older time they be, living few and secretly, wild and wary as beasts."
Acarin
Posts: 202
Joined: Fri Sep 04, 2020 10:49 pm

Re: Cogg demographics

Post by Acarin »

Leatherworking may be unpopular because of the heavy skill investment. I suspect that you would need high tracking, skinning, and leatherworking to make decent quality armor. People in a profession typically prefer to be able to gather ingredients and make the merchandise themselves so it's a heavy investment. I would take those skills as my next top choice (after combat) if I had the points to invest.
You reach toward ((DEV Rias)) ... Pull(d225([1]x)):214 vs Mark(d1100):714 = -500 (-222%)
You notice ((DEV Rias)) glance your way, causing you to quickly withdraw your hand from his wool
drawstring pouch (open).
Roundtime: 5 seconds.
User avatar
Lexx416
Posts: 563
Joined: Wed Sep 06, 2017 6:14 pm

Re: Cogg demographics

Post by Lexx416 »

Acarin wrote: Wed Oct 28, 2020 9:07 pm Leatherworking may be unpopular because of the heavy skill investment. I suspect that you would need high tracking, skinning, and leatherworking to make decent quality armor. People in a profession typically prefer to be able to gather ingredients and make the merchandise themselves so it's a heavy investment. I would take those skills as my next top choice (after combat) if I had the points to invest.
In regards to skill investment, it depends. I have 200 tracking and I have no problem finding game. 200 tracking , 200 stealth, 200 ranged, and 100 perception and I can easily deal with almost all of the game I've encountered, except for maybe coyotes. Those are hit or miss sometimes because of their perception. I only have 25 skinning, so I can't get high quality pelts (unsure if that's even a thing right now??), but I've also never gotten anything worse than an average pelt.

Obviously doesn't take into account more dangerous animals, or rarer pelts.
"You hear the Woses, the Wild Men of the Woods... Remnants of an older time they be, living few and secretly, wild and wary as beasts."
User avatar
nobody
Posts: 501
Joined: Wed Jun 17, 2020 7:43 am
Contact:

Re: Cogg demographics

Post by nobody »

Alright, today I wanted to finally tackle the guild/class data.

Vanillas are quitters: I wanted to start off addressing the characters that haven't yet signed up for a class. That's 208 guild-less, 8 vanilla warriors, 6 vanilla adventurers, and 2 vanilla scholars. What all of these characters have in common is that they belong to a low-level cohort. The average level for the vanilla warriors, scholars, and adventurers is 1.4, 2.0, and 2.2. Because the counts are so low (8, 2, and 6 respectively), we can conclude that they are all low level. In particular, the highest possible level for a vanilla warrior from this set is 4, 3 for scholar, and 8 for adventurer. Probably they are all level 1-3 though. The 208 guild-less characters have a similar average level of 2.1, but because there are soooo many of them, it's likely there is at least one high level character hidden in the lot, waiting for the day when they can claim their artisan guild and shine. The rest are at best people that have tried the game and left it or test characters, and at worst they are name reservations. Bleh, name reservations. Incidentally, if the introduce command that most people seem to prefer not using were removed, there would be little practical reason to have a limited name reservation system. I know more than one David, I get by just fine. I can also see all of the visual distinctions between them, but VI people can also differentiate without that tool. But that is a side ramble for another post perhaps. Onto the classy folks!

Measures of popularity: There are a few ways to try to tackle the question "what's most popular?" The counts suggest the top three are physicker, ranger, warlock and the bottom are rogue, marauder, and duelist. The ranking by level is similar: physicker, warlock, ranger on top, guardian, rogue, marauder on bottom. And by total experience the ranger falls away from the top, giving us: physicker, primalist, warlock on top and guardian, rogue, marauder on bottom. Those aren't the most interesting for me though. Average level (total level divided by total count) changes things up subtantially (but only for the top): arcanist, dreadnaught, duelist are now on top but the bottom is guardian, rogue, marauder. We can also reverse engineer another average level type measure by taking average experience and converting that into levels and gives the same set, but re-ordered: dreadnought, duelist, arcanist for the top, but the bottom remains guardian, rogue, marauder. So what's most popular? A bunch of people try physickers, ranger, and warlock. Some people play dreadnought, duelist, and arcanist but they tend to play them more, giving a higher average level. What are people not playing? Guardians, rogues, and marauders are definitely on the bottom. Guardians kind of make sense, we still have limited numbers and guardians are going to shine the most in a group and defense first is just going to be a slower haul compared to dreadnoughts more offensive approach (pun intended!). Rogues are about half as popular as treasure hunters, which right now are mechanically very similar (though not identical if the wiki is to be believed). This also makes sense - if you want to play someone who goes out and collects treasure, you can. Great. If you want to play someone that is sneaky and into all manner of back-stabbery, you might need more people around. Unless you just want to rob Hoss all the time - I guess some people think that's fun. Marauders are the outlier perhaps, as they don't shine more with more people, they may just be less developed or otherwise less mechanically interesting. Perhaps all the marauder needs to shine is for someone to pick up their torch and become their champion for a time.

Finding champions: The average xp derived level is ALWAYS at least as high as the average level because maths and non-fractional levels. That means there is a measurable gap between the two. But does it mean anything? I speculate that it does: Arcanist, physicker, duelist, dreadnought, and primalist all have a gap greater than 4 (4.5-6.7). Marauder, guardian, berserker, bard, treasure hunter, rogue, ranger, and nightblade all have a gap less than 2 (0.7-1.5). Warlock is perhaps between the two groups at 2.4. I believe the gap might be a measure of play time or whether or not the class has found a champion, but perhaps time will tell.

Guilds: Interestingly, when lumped into guilds ordering becomes very easy. Ranking by count is scholars, warriors, adventurers. Ranking by level is the same, ranking by total XP is the same, ranking by average level or average xp are also the same. Wait, did I say interestingly? I lied. Very boring.
User avatar
Rias
DEV
Posts: 2024
Joined: Sun Sep 03, 2017 4:06 pm
Location: Wandering Temicotli

Re: Cogg demographics

Post by Rias »

Always interesting to see some insights on the data. I love the info, but I'm too lazy to do this kind of analysis myself, so thanks for sharing your findings!
<Rias> PUT ON PANTS
<Fellborn> NO
User avatar
nobody
Posts: 501
Joined: Wed Jun 17, 2020 7:43 am
Contact:

Re: Cogg demographics

Post by nobody »

I get that. Also, I'd much rather your time be spent tackling your to-do list of awesome or showing up and making the world feel alive instead of analyzing numbers. Thanks for making them available so I can do some of that lifting instead ;)
User avatar
Rias
DEV
Posts: 2024
Joined: Sun Sep 03, 2017 4:06 pm
Location: Wandering Temicotli

Re: Cogg demographics

Post by Rias »

nobody wrote: Mon Nov 02, 2020 10:31 am and at worst they are name reservations. Bleh, name reservations. Incidentally, if the introduce command that most people seem to prefer not using were removed, there would be little practical reason to have a limited name reservation system.
Incidentally: You can use the NAME command on yourself. So if you decided you wanted to start going by Franklin instead of Bob, just NAME SELF AS FRANKLIN. The INTRODUCE command uses whatever name you've given yourself, not your hardcoded character name. (Of course, this won't affect what other people have already named you from their perspective.)

So there's not really any point in making characters just to save names, unless it will -really- bother you that much to see it on the login screen, since you can just name yourself however you please at any time, and just tell other people your name is whatever you want it to be.

Er, and fishing records, I suppose. Those use your hardcoded name. Could easily update that, though.
<Rias> PUT ON PANTS
<Fellborn> NO
Post Reply