Tracking exploration activity

[marx]

I've been tracking data about exploration activity through EDSM for some time now, and finally wrote things up and put them in a public spreadsheet to share them. Since the additions of squadrons and leaderboards, there's also additional in-game data to track.
Credit goes to EDSM for the data of course, and Orvidius's EDAstro helps me with not having to run planet counts on the whole database (or keep an updating local copy of it) every month. Additional credit goes to Qohen Leth for the DW2 statistics, which will be referenced below, and to Ian Doncaster for historic squadron leaderboard data.

Now, here's the data with charts, and in the notes, the dates of events which I thought were of significance:
https://docs.google.com/spreadsheets/d/1yw7olxlZUJIqYbrP3jbE3eBccIwuwVMo45qTVhZOy7Y/edit?usp=sharing
There's also a collection of one-time snapshots and data which I no longer track, you can find those here:
https://docs.google.com/spreadsheets/d/1mERFq6z1J_RRpauXCKdCQ56Iv5SfEPD4KBPrHaA4n30/edit?usp=sharing


For an analysis on how representative this all might be, see at the end under the spoiler cut.
If anyone has suggestions for more events of significance to add, let me know!

For the record, planet counts exclude the stars. (That's why I didn't call them bodies, as the game puts them both under the same category.) This makes it easier to distinguish between the auto-scanned stuff and what (currently) needs to be done manually.

I'll post my thoughts to date in a second post below.

As mentioned above, the statistics:

Spoiler

During DW2, 13,608 people signed up, 5,380 of them on EDSM. That's 39.53%, much more than enough. At the end of DW2, 3,747 people finished, 2,628 of them on EDSM. That's 70.13%. Which means, interestingly enough, that if you finished DW2 and weren't on EDSM, you were in the minority.

Let's go with what we know from Frontier then. On 2018. March 2, they wrote that a total of 112,863,791 systems were discovered. Going from the 2014. Nov. 22 Gamma (the head-start) release date, that's an average of 94,367 systems added daily.
EDSM started logging systems on 2015. May 15. However, back then, it could log far less systems, as they had to be trilaterated until 2016. February 1, meaning that players added far less systems manually than they have actually been to. Now, on 2018. March 2 mentioned above, there were 20,853,341 systems added, making it 20,345 per day.
Extended to the present day, it would be 28,374 systems per day - but we don't have the current statistic from Frontier. Maybe somebody could ask them?
Also, to better put things into perspective, during DW2, the average on EDSM was 61.6k per day. (That also includes the months when numbers started dropping, as many finished it before the official deadline in the middle of June.)

But let's move back to 2018 then. So, what we do know is that EDSM had an average of 20,345 system uploaded per day, while the entire game across all platforms was 94,367. (Bear in mind that at this time, EDSM had no console support.)
So, if we're comparing systems, then EDSM represented 21.56% of the total. Plenty good. I think we can safely say that a player who doesn't upload to EDSM doesn't explore significantly slower nor faster than a player who does, so we can use the same speed for both groups, and then the ratio should remain the same.

That's a relative sample size, what about an absolute then?

To do that, we'll have to estimate how many systems per day a player would explore on average. Now, looking into the DW2 statistics that Qohen Leth posted and the EDSM data, we can estimate that. I went with a low estimate of only 2,628 DW2 players uploading their finds, but that could certainly be refined, by comparing daily statistics with how many people reached each waypoint, and so on.
But for now, the per-player average would be 10.12 systems a day, while also scanning 108.86 bodies, plus auto-scanning 18.42 stars. So let's go with 10.12 systems then.

Again, we can say that a player who doesn't upload to EDSM doesn't explore significantly slower nor faster than a player who does, so we can use the same speed for both groups. With that, we have a daily average of 9,325 explorers, rounded up. Out of them, 2,011 would be on EDSM. (Comparing these to the exploration CG contributor statistics, which probably don't really attract many explorers, and Steam player counts, which are only a part of the PC player count, both sound plausible.)

So, for a standard 95% confidence level then, with those population and sample sizes, the margin of error would be 1.93%. That's plenty good.

As a later update showing per-year charts and summarising what happened in each year, from 2017 to 2021, see this post.

 

[marx]

Let's see then. Personally, I think it's quite interesting that for the most part, activity has hovered around 1,000,000 new systems monthly, and has returned to this level after some events. (Let's not forget though that EDSM added console support soon after Beyond Chapter Four.) For the effects of various releases, let's see:

• Both the Return and Beyond Chapter One brought a significant boost, but two-three months later, this went back to the "baseline".
• The minor updates of Chapter Two and Three had seen a slow decline to below the "baseline" activity, and then...
• The reveal of the FSS and especially its reveal on the livestream resulted in the lowest activity to date, in 2018. November.
• The launch of Chapter Four and the introduction of the FSS caused a big spike, and the launch of DW2 resulted in an even larger one. (Note that this is best seen if you plot a daily graph, not a monthly one.)
• Overall activity in both systems and bodies scanned peaked during 2019. February, before DW2 reached Sagittarius A*. Scans started dropping quickly afterwards, while systems followed suit a couple of months later. People still discovered as many systems, but used the FSS significantly less.
• DW2's end lasted almost two months, and this is easily seen on all charts.
• The April and September updates (of 2019) don't seem to show up at all, one way or another.

What I didn't really expect was the ELWs / Systems change. From a slow but steady increase for 1.5 years, it started dropping with Chapter Four already, and was halved by the time DW2's end came. Since ELWs are the most likely bodies to be scanned, I think this shows more people chain-boosting neutron stars, in such a hurry that they didn't even notice ELWs.
Another possible reason could be that it's easier to mix up ELWs with rocky ice worlds though, and that's why I think the same decrease is markedly less when we look at AWs / Systems.

Another thing that I didn't expect was that Stars / Systems would still change after the auto-scan was added. My guess is that the increase up to 2.07 was because the majority were inside the core (on DW2) then, and we're back to around 1.5 because explorers are more spread out in the galaxy now.

Oh, and for a long while, people have debated why EDSM had more ELWs uploaded to it than AWs, while area surveys done by people pretty much all showed that the latter are more numerous. With the FSS having made recognizing and scanning ammonia worlds much easier, this question was settled.

Overall though, it seems that despite the reward shower of the Chapter Four update, it has failed to make a lasting impact in activity. On the bright side, it did mean that more stars and planets are being uploaded, but that's due to its design, and not to people exploring more. So I think we can at least safely say that in general, the people who upload to EDSM aren't primarily in it for credits and tags

As for the April and September updates, they don't even show neither here, nor on the daily stats. Of course, they were even smaller updates than Beyond Chapters Two and Three, and since they were geared towards new players anyway, that's hardly a surprise.

 

[Redfox]

Nice write-up, thanks CMDR!

Found myself agreeing most of you conclusions - especially the ELW/AW thing. I think we all assumed the reason, but it's nice to finally have some 'confirmation' that it actually did come down to the AW on average being a bit more difficult to identfy on the SysMap than ELW.

One little nitpicking comment on the 'historical' bit in the spoiler.

However, back then, it could log far less systems, as they had to be trilaterated until 2016. February 1, meaning that players added far less systems manually than they have actually been to.

If I recall correctly: The log files gave us the system name pretty early on (remember having to set verbose logging after every ED update and forgetting half the times? ), to be scraped, stored and eventually submitted. Thus, even back then the systems as such were available in EDSM (or was it still EDDB?), irrespective of its missing or trilaterated xyz-position or their missing bodies.

I think the only time when "players added far less systems manually than they have actually been to" was before system name became available in the verbose jump-event in the log file, up to then requiring us to log the system names manually - but TBH, that's where my memory gets fuzzy (probably denial of the time spent doing that).

In any case, bless FD for the eventual introduction of the journal files!

 

[Orvidius]

As an aside, my EDSM feed is showing 2.6 million systems still missing coordinates. I checked with both "and" and "or" booleans, and the number is thankfully the same:

mysql> select count(*) from systems where coord_x is null or coord_y is null or coord_z is null;
+----------+
| count(*) |
+----------+
|  2650371 |
+----------+
1 row in set (4.83 sec)

mysql> select count(*) from systems where coord_x is null and coord_y is null and coord_z is null;    
+----------+
| count(*) |
+----------+
|  2650371 |
+----------+
1 row in set (1.59 sec)

I have no idea how many of these might actually be duplicates in EDSM's data though. I suspect some data gets orphaned when FDev renames a system or something, but that shouldn't account for too large a number.

 

[marx]

If I recall correctly: The log files gave us the system name pretty early on (remember having to set verbose logging after every ED update and forgetting half the times? ), to be scraped, stored and eventually submitted. Thus, even back then the systems as such were available in EDSM (or was it still EDDB?), irrespective of its missing or trilaterated xyz-position or their missing bodies.

Oh yeah, I had actually forgotten this. Could also be because such systems that were missing their coordinates didn't show up in the videos. Thanks for reminding me!

As an aside, my EDSM feed is showing 2.6 million systems still missing coordinates. [...] I have no idea how many of these might actually be duplicates in EDSM's data though. I suspect some data gets orphaned when FDev renames a system or something, but that shouldn't account for too large a number.

That sounds roughly about the same amount of systems that were added the "big explosion". Add to that client and/or upload errors, and... it's still more than I would expect. They're probably all places that nobody visited again since. Out of curiosity, could you put them on a map? I have a suspicion that many of them would be in high-density areas, the core.

 

[schlowi123]

Thank you for this analysis. Interesting to read, as always. Can't add anything to it, partly also because I play on console.

And this …

[…] I checked with both "and" and "or" booleans, and the number is thankfully the same […]

… made me laugh really hard … the joys of programming

 

[marx]

Thank you for this analysis. Interesting to read, as always. Can't add anything to it, partly also because I play on console.

Actually, I wonder about something: how difficult is it in practice for a console user to regularly upload systems to EDSM? It would be good to know whether it's more difficult than the streamlined process(es) on PC or not. I'm wondering about how many "hidden" systems there might have been before console support was added.

After all, what we do know is that 19% of those who finished DW2 did so on consoles, but not how many of them were also on EDSM. But let's suppose that 50% of console users who finished were on EDSM too: in that case, their systems would nearly account for 10% of the total. If that ratio were true, then the actual pre-console support (before 2018. Dec.) system numbers would have been 10% higher.
But if it's still difficult to upload all your finds there, I'd imagine the real numbers would be much less than that.

 

[Orvidius]

Out of curiosity, could you put them on a map? I have a suspicion that many of them would be in high-density areas, the core.

Not really, since the coordinates are missing.

I could probably create a map of sectors colorized by how many have that sector name, or something.

EDIT: I also have a spreadsheet of these: https://edastro.com/mapcharts/files/systems-without-coordinates.csv

 

[marx]

Not really, since the coordinates are missing.

...
...
I derped

But hey, thanks! By a quick look, they're mostly familiar sector names, plus I forgot about the newly-locked sectors. I'll do a count later and throw the results at the sector map here, we'll see if some pattern would emerge.

 

[iain666]

Not really, since the coordinates are missing.

There are the estimated coordinates from system name which are generally good to +/- 15ly or so but last time I checked the EDSM API doesn't expose that so you'd have to re-implement. Which I wouldn't want to do so wouldn't expect you to.

 

[marx]

Hm, it'll take me some time to do that sector map instead, as I won't have the time for it for a few days.

Also, I realised that I made a mistake: the "baseline" would have been somewhere above the one million systems mark, if it was one million before console support was added. In that sense, we likely haven't returned to that (yet?), but by how much is an open question until we can estimate just how much console players contribute to new systems uploaded now.

 

[Ian Doncaster]

I did check this once and they're broadly distributed similarly to where people were trilaterating before 2.1 released - so lots in the core, bubble, DW1 route, few elsewhere. Some of the ones nearer to the bubble are of course now rediscovered - so there are more in Praea Euq than Wredguia, for example.

 

[Orvidius]

There are the estimated coordinates from system name which are generally good to +/- 15ly or so but last time I checked the EDSM API doesn't expose that so you'd have to re-implement. Which I wouldn't want to do so wouldn't expect you to.

I've been on the fence about whether to take a stab at that. I finally decided to give it a shot. It's possible to get fairly precise location estimates for low mass codes, since their subsector sizes are small, but the larger mass codes can be located anywhere in a much larger cube. So I'm drawing rectangles that are sized based on the area that each system could be located in. And as it turns out, the systems are all over the galaxy, as we might expect:

Spoiler

 

[iain666]

I've been on the fence about whether to take a stab at that. I finally decided to give it a shot. It's possible to get fairly precise location estimates for low mass codes, since their subsector sizes are small, but the larger mass codes can be located anywhere in a much larger cube. So I'm drawing rectangles that are sized based on the area that each system could be located in. And as it turns out, the systems are all over the galaxy, as we might expect:

Spoiler

Less than 3m left to trilaterate, 20 at a time

 

[Orvidius]

Less than 3m left to trilaterate, 20 at a time

That's a pretty terrifying thought...

 

[schlowi123]

Actually, I wonder about something: how difficult is it in practice for a console user to regularly upload systems to EDSM?

Not easy at all. One needs to log in at EDSM and then open the dashboard. Once every minute the pilots history is downloaded from Frontier. Albeit the browser needs to be on all the time. In general the dashboard tab does not need to be the active one, but after a while EDSM seems to forget that it shall load my discoveries from Frontier. Well, reloading the tab helps with that.

The good thing is, I don't need to play for that. I can just open the dashboard and EDSM does what it does even if I open the dashboard at work.
When I had forgotten to update for a couple of days the backlog can take quite some time to be downloaded. Especially if many jumps took place or discoveries were made or just docking at stations, it can take a while before the activities of the busy day are updated.

Anyway, opening the dashboard is easy as 3.1415… and since everything else is done automatically that can't be considered complicated either.

BUT … I've heard every now and then getting this whole thing to work at all can be difficult. It worked like a charm for me but I don't know how many console players aren't doing it because of this intitial step.

 

[Orvidius]

I made a minor correction to the missing coordinates map. The vertical banding was due to mixing up X and Y in one of the equations. It's a little clearer now:

https://edastro.com/mapcharts/missing-coordinates-map.jpg

(not including it inline, since the forum has cached the old one)

I've also added it to the website.

EDIT: I'm also adjusting the color ranges to give a little better fidelity too.

 

[marx]

A question occurred to me: when people explore more systems, do they still scan the same average of bodies in them, or less, or perhaps more? The daily data should help here, because the weekends always see more players in games. As such, here's the number of systems discovered daily, and the bodies per system (multiplied by 1E4, which is nothing more than an arbitrary value so I can put them on the same chart on Google Sheets):

And to visualize things a bit better (but not terribly useful), the same with polynomial regression (fourth degree) added:


I think it looks interesting. You can see on this one as well how players started to lose interest in scanning in February even while they still went to the record amounts of systems, and with more activity during DW2, the "amplitude" of the weekly swings was higher than they are now. Oh, and unlike the monthly one, you can see the effect of the Explorer's Anchorage building CGs on this chart.
Speaking of which, while the April Update's launch day isn't easily visible (remember, DW2), the September Update's is.

But back to the question, of how scanning habits might change with more systems visited: this is in no way an accurate analysis, but I think there is something there. On a short term basis, when systems go up, bodies / systems often goes down, and vice versa. This could be due to a number of reasons, many involving playing habits.
And well, there aren't really any practical implications or uses of this (or at least, none that I see), but I'd say it's still a nice curiosity

 

[schlowi123]

Regarding the practical use of something: Curiosity might kill the cat but it certainly brought us to the moon … and before that over the oceans … and before that fire … and before that down from the trees … and before that up on the trees … I don't think anoybody found that really useful at that time.
I nyself find those real patterns (and possible explanations) super interesting. Much more than in the data from ED objects where we can see all kinds of patterns and it all comes down to "because that's how it was programmed" … so … please carry on with you analyses

 

[marx]

A smaller update to this: I've been looking at the official data from squadron exploration leaderboard scores (which is roughly equivalent to the credit value of exploration payouts) and should have plenty to post on that soon, but to better separate things, I'll post a smaller part here. To better estimate how many explorers might play on console, I asked for the top 10 ranks on all three platforms. Thanks to @Ian Doncaster, @schlowi123 and @FixMeNow, here are the results (the full data will be at the end, in case anybody wants it):

Season 6 Top 10 Total	212,797,160,328
PC	73.12%
PS4	17.50%
Xbox	9.38%

Also, the standings of the current season as of Oct. 23:

Season 7 in progress	22,038,981,061
PC	74.82%
PS4	10.21%
Xbox	14.97%

Compare this with the DW2 statistics, where the consoles made up 21.67% of the sign-ups and 18.82% of the arrivals.
Personally, I was surprised, as I expected the ratios to be less, not more. Of course, bear in mind that this is squadron data, and AFAIK we have absolutely no idea about how many explorers are in squadrons and how many outside of them. Still, I think we can estimate that somewhere around 20-25% of explorers play on consoles then.

In that case though, if we go with console explorers exploring at the same speeds as PC players do, and then "correct" the pre-Chapter Four numbers with +25% (EDSM added console support a week after the Chapter Four release), then things look worse than they did before Chapter Four. Not by a lot, but still noticeable. The lowest point then still came after the FSS's plans were revealed, but when it comes to new systems at least, things looked better before than after, and likely more people explored then than they do now. Except during DW2, that is.

Of course, we'll see how things go from here, what with Frontier announcing a content freeze until next summer, and seeing if they'll fix the many exploration bugs and design issues in the upcoming fix patch(es), or not.

Next up: data (thanks again to Ian!) and analysis on the previous PC season leaderboards, beyond the top 10 too.


The full data then:

Spoiler

PC Season 6	PS4 Season 6	XBox Season 6
50,205,452,385	10,645,383,760	6,802,493,218
33,438,560,270	7,834,759,766	4,714,564,198
13,312,253,671	5,384,627,873	1,385,947,852
12,761,707,669	4,890,199,907	1,339,909,790
9,126,466,246	2,208,045,166	1,236,572,909
8,329,602,967	1,337,012,713	986,834,518
8,101,260,866	1,326,689,206	956,087,229
7,044,934,847	1,322,160,311	904,140,861
6,717,568,874	1,141,871,119	817,657,067
6,568,135,178	1,138,678,431	817,581,461

PC Season 7 WIP (Oct. 23)	PS4 Season 7 WIP (Oct. 23)	XBox Season 7 WIP (Oct. 23)
4,627,064,178	401,505,508	1,203,776,286
2,961,585,090	264,411,874	715,674,596
2,955,144,725	263,620,685	461,446,912
1,524,139,597	217,882,493	172,651,558
1,361,335,286	200,551,138	133,337,057
719,501,046	199,145,174	127,887,086
709,624,131	196,839,614	125,163,722
640,323,106	177,746,501	124,273,389
501,611,056	173,388,753	120,385,216
489,054,357	154,300,680	115,610,247