Hi everybody! I was playing around with the EMC website API and collecting some data for fun. I decided to keep a record of staff activity levels, and after one week of data, here is what I have found. I polled once every 5 minutes, however because the polling isn't instant, it's technically slightly slower than every 5 minutes. I could have changed it to be exactly every 5 minutes but it wouldn't change much with the data for now. How often are staff on? According to my data, staff are on 40.7% of the time. This does not account for multiple staff on, this is just how often there is at least one staff member of any rank online. Here's a time graph of staff activity during the day This represents the average number of staff on per hour. Unfortunately, the axis labels aren't extremely visible the way they were attached to the picture, but the notched on the left represent 20%. Each hour has it's on spot on the graph from 0 to 24, EMC time. The graph goes from 0% to 160%, as this graph is weighted by how many staff are online. (If 2 staff are on, it counts as twice as much.) Here's the same graph, but unweighted. That means that this simply represents the average amount of time which has at least one staff on per hour. Once again, it's 20% per tick on the side, 24 hours EMC time. How long might one have to wait for a staff member? I tracked the streaks of time where there was no staff. Here's a graph of the streaks, grouped by 5. Grouping by 5 made the data a little nicer. The notches on the bottom represent 20, and on the side they represent 5 "polling instances" where there was no staff on. Here's the raw data, ordered: {0: 30, 5: 11, 10: 8, 15: 7, 20: 3, 25: 5, 30: 2, 35: 1, 40: 1, 45: 2, 50: 1, 55: 1, 60: 1, 65: 1, 90: 1} The streaks are rounded down: 0 means 1-4, 5 means 5-9, 10 means 10-14, so on. The streaks represent polling instances, which are 5 minutes apart, so simply multiple the number by 5 to get the actual duration. The one 90 length streak means there was a a length of almost 6 hours where there was no staff on. Here's the individual staff activity during this week (The number represents how many times the staff was online when the program polled the API) { 'Luckygreenbird': 6, 'deadmon5': 6, 'BurgerKnight': 11, 'AlexC__': 13, 'Krysyy': 16, 'Hashhog': 23, 'weeh': 26 'carolmoss': 51, 'MoreMoople': 70, 'jaqque': 73, 'AncientTower': 84, 'NickkG': 94, 'Burki': 138, 'AnonReturns': 207, } Any staff who don't appear in this list weren't online at all for that week. This data certainly makes a good case for anon and burki getting a promotion Here is a link to the raw data for the statisticians on EMC who want to play with it : https://pastebin.com/3jS2Vc3a I am continuing to collect data, I hope to run it for a month and then re-run these same statistics. One issue with the data currently is that there is a 12 hour gap in the middle of it due to the server I was running my program on having issues. I migrated it to a much more stable one so hopefully no issues with that anymore. So, next time you are looking for staff help, make sure you're on during one of those spikes in the graph 5:00 PM seems to be your best bet! Also, remember, this data is very rough and doesn’t account for a lot of possibilities such as vanished staff, or staff doing online work. So don’t take this as a definitive judgement on who’s doing work and who isn’t.
would it be possible to do one of these for All players too? (i know there would be afk players in the data but it would still be interesting Very nice data though
This only counts for staff on the smps and the games server. I wanted to do the stage server but was having trouble getting the server ID to be able to get that data.
Yes, the API gives all players on a server and they each have a number that represents their rank (0 = guest in tutorial, 1= normal player, 2,3,4,5 = donators, 6 and above are the various staff ranks). I only took down player names that had 6 or higher, but you could easily do the same thing for all players. However, the data would get much bigger much faster and I wasn’t super interested in it myself, although I could pretty easily add a player count to it to just record how many players and not their name.
ah cool, would be interesting but is there a way to reasonable track more? so that we could see which normal players play the most?
Yes, but that would require keeping records of all the players. Though, come to think of it, there’s probably a better way of doing that then the way I’m doing it with the staff that might be more space efficient. I’ll look into it.
That's cool! It's been a while since I saw something so interesting on EMC. It would seem better if staff members were on most of the time, right?
It’s tough to increase activity among existing staff members, perhaps making more ranks off staff and giving promotions as rewards for activity. Also, hiring more staff can help compensate as well. If you look at the list of all of the players who are technically staff, a very small fraction of them actually appeared even once in that one week span.
I'd like to point out a major flaw in your data. While I get that you were doing this to ascertain which staff were online in-game, the staff are using Slack/Square/Track and sometimes on alts in-game instead. A simple /report pings the slack channel and players' need to wait for staff is actually MUCH lower than represented in your data. As discussed, the staff server 'stage' is also not in your data. October event final testing/setup is underway and so we have a lot of activity there that you can't see with this poll. Your data on Hashhog alone proves that to me because they have been doing extreme work to prepare and work with build team, yet they are not extremely active in your charts. In summary, I've reviewed your data, but it doesn't match mine. So for anyone looking at this, please keep that in mind. Also, activity is not the primary reason behind promotions because the difference between moderators and senior staff is not defined in such a way. There's so much more to it than that and we'll continue to operate in the same fashion as we have going forward. If I seem defensive, just know that I take the time to defend the activity of the staff, because I know how hard they work behind the scenes to keep this place operating smoothly. Simple numbers polling from an incomplete data set should not be used in any manner to judge any of them as this post is doing by listing them out. And yes, I understand that you are doing this for fun, but I needed to define the holes present before anyone used your data to pass judgement on the staff team in any way.
Absolutely. By no means is the data perfect, it also doesn’t account for time spent vanished, nor for work done on the forums or anywhere not in game. 5 minutes is also a fairly large gap between polling, but I go that slow because then I only average 2 requests per minute which is about as fast as I want to go. Someone could log on and be completely missed by the program if they vanish right away or whatnot. I’ll add a note to the original post making sure that’s clear.
