Analysing the ELW data from EDDB/EDSM

I set out to look into the distribution of Earth-like worlds in the galaxy, to see if there are any regions where they are more frequent. I got the data for this from the EDDB and EDSM nightly dumps (I worked with the Aug. 26 dump), so the first thanks go to their respective teams: this much data wouldn't have been possible without them. Then, iain666 and Redfox helped me with a program to build an SQL database from the bodies JSON, so the second round of thanks go to them: that saved me a huge amount of work.
Thanks also to Ian Doncaster, for a bit of Unix help so that I could get the total systems calculation run more swiftly.
Lastly, for the newest versions (2019.05.30), thanks a lot to Orvidius as well! Those automated spreadsheets saved me a lot of work.

Rather than discard the characteristics (not really needed for the distribution), I instead them as well, to go over them later. So I compiled the characteristics of the Earth-likes and did some analysis on those too.
Before I go into things further, here's the data in two separate sheets:

The current version for 2019/05/30 is at
An additional version containing separate data and charts for ringed ELWs:
An earlier version with ringed ELWs for 2018/05/30 is at
The older version from 2017/08/26 is at

Further analysis on per-ELW breakdowns for 2019/05/30 is at Note that this doesn't include the ELW data itself, just the results: Google Sheets couldn't handle that many data. It does, however, include ringed ELW data. If somebody would like the data too, send me a PM and I'll send it over.
An earlier per-ELW breakdown for 2018/05/30 is at It contains coordinates, and rudimentary map-charts rendered from them.
The full detailed planetary data (with statistics) for 2017/08/26 is at

I've removed terraformed Earth-likes from both lists, as well as ones in non-procedural star systems. Those might still be created with the regular Forge generation, but I wanted to be sure.
Also, there were some ELWs with value errors: I removed those from the second list, but not the first.

Missing is any information on rings, and moons aren't included either. Perhaps those will come later. But first, I'll look into how things differ in different mass codes (if they do at all).

About the distribution of Earth-likes:
The maps on this sheet are based on EfilOne's masscode sectors map.
To me, it looks like we don't have enough data for anything conclusive yet. From what we do have though, I'd say that significant differences between regions are unlikely. When looking at the ratios between known ELWs and known systems, sectors of special interest stand out, especially the places where players can dock and sell their data. Just take a look at the Eol Prou sector. (If the vicinity of the bubble weren't carved up with all those non-procedural sector overrides, it would likely be the same.)
Basically, the places stand out where players have searched for Earth-likes or done detailed surveys, as opposed to areas where most people just travelled through. Of course, the Sol-Sag. A*-Jaques lines are still significant.

Also, the ratio is only calculated for sectors which have more than ten known ELWs and one thousand known systems. You can easily adjust both values in the top row (if you make a copy of the sheet), and experiment with how things would look like with other limits.

Rather than looking at the entire galaxy, I think it would be better to look at thorough surveys of specific sectors, in different regions of the galaxy. (Some near the edges, others in the core, and so on.)

About the characteristics of Earth-likes:
This was in part inspired by Sapyx's recent thread. It's interesting to compare how things differ on this larger sample.

I included only two histograms as examples, since Google Sheets can't really run more with this much data, but it's interesting to see others too. You can easily modify them to show others: just change the range.

Seems like smaller Earth-likes happen to be more popular with the Stellar Forge. Especially interesting is how the universe seems to hate ELWs with gravities around 0.86-0.92 g. Likely an effect of how they are generated - and I wonder if there are two (or more) "types" of how this happens. I'll try to see if this might be due to differences between mass codes.

Speaking of gravity, there doesn't appear to be a hard limit for it (like for temperatures): instead, it's likely controlled indirectly by other variables. Namely that the atmosphere has to be breathable.

Also, speaking of the Forge hating things, note how rare eccentric orbits are for ELWs. Little wonder though.

As for volcanism, there were six Earth-likes with nitrogen volcanism, and one with water magma volcanism. I'd check these for possible bugs, especially the water one. Sadly, it's not on the ELW list.

A bit of trivia: the farthest distance a Commander had to fly to an ELW was 660k ls. Right on, Commander Goemon!
As I mentioned, I'll look into more things later. Right now, I'm looking at doing subsets based on mass codes, and separating Earth-like moons too. Once I do and find anything interesting, I'll post it in this thread again.
Thanks for reading! If you spot any errors, do let me know!
Last edited:
I've included a check for Earth-like moons, and made a separate listing of those. Some mildly interesting things about them:
There is one ELM in a mass code B system, while every ELW in mass code H systems (all twenty of them) are ELMs. Only sixteen ELMs have any volcanism, but they have four out of the six ELWs with nitrogen magma volcanism, as well as the lone water magma volcanism.

Bear in mind that there are two distinct types of Earth-like moons: those that orbit a planet, and those that orbit stars. By the latter, there aren't just those that orbit L,T,Y dwarfs, but those that orbit more luminous others. For example, a "classic" configuration of the above would be an ELM orbiting a class M dwarf that's not a primary in a system, but is orbiting a black hole instead. Hence why all the ELWs in the huge mass code H systems are moons.
Of course, looking at the statistics, all the moons are lighter on the whole. Not surprising considering that more of them orbit bodies than stars.

Also bear in mind that the surface temperatures of atmospheric moons are bugged, and once that's fixed, their temperatures will rise. As such, any ELM you find might turn into a WWTC later. Although I wouldn't be worried about this any time soon, as I expect FD will fix this bug only when they'll revisit atmospheres, when landing on some atmospheric planets comes.
Fantastic work, marx - as we've come to expect ;)

Any chance to generate a map with the ratio of amount-of-ELW per amount-of-known-systems per sector? I guess/hope this would reduce/eliminate the bias of highly visited sectors.
@ Redfox: Thanks! And sure, the first linked sheet has that already, see the "ELW ratio" subsheets on it. You can also tweak the cut-offs on the "Data" subsheet. (I've set them as minimum 10 Earth-likes and minimum 1000 systems per sector, so that rarely-visited places wouldn't produce ratios all over the place.)
Last edited:
"... the farthest distance a Commander had to fly to an ELW was 660k ls. Right on, Commander Goemon!" - when was this (beside "while preparing dinner") and where?

am i finally a record holder? :D

interesting read as all your threads, thanks for sharing!
"... the farthest distance a Commander had to fly to an ELW was 660k ls. Right on, Commander Goemon!" - when was this (beside "while preparing dinner") and where?

am i finally a record holder? :D
Smojeia ML-Y d3, April 4 this year. Although we don't know if it was tagged by you or not, and if it was you who flew that distance, or the nameless other Commander who has been to the system too. So, who was it first?
Smojeia ML-Y d3, April 4 this year. Although we don't know if it was tagged by you or not, and if it was you who flew that distance, or the nameless other Commander who has been to the system too. So, who was it first?
oh yes, that was on the way back from HT sagittae, an impressive supergiant.

system has 2 WW and an AW around the main star, an ELW around the secondary additionally to those hit me as unusual, and anyway i go scanning any ELW.

if we can ever land on those planettypes, that system is on my list of places to visit.
Excellent, thanks for the follow-up study. Some comparative observations of my own:

Mass: Not too much difference between my subsample and your macro study, there, though both averages are lower for yours.

Gravity: The Earth-matching averages have disappeared, as one would expect given the lower mass averages.

Temperature: The averages are 2 degrees higher, but interestingly, the two degree separation between the two averages is preserved.

Atmospheric pressure: Your data almost exactly matches my observations here.

Day length: These two averages and the wide difference between them also match up nicely.

Volcanism: 15.6% of the worlds are silicate magma; a lower percentage, but expected if lower-mass planets are more common than I'd observed. One water-magma and six nitrogen-magma ELWs is an impressive minority; that makes them about as rare as Glowing Green Giants. I do note that there isn't a single Ammonia Magma ELW found yet, yet I have found at least one Ammonia Magma terraformable. I'm assuming that "terraforming" does not include large-scale re-engineering of a planet's basic chemical makeup, so if terraformables exist, then an ELW might exist out there too. I for one will keep looking.

Maximum ELW-matching points: I am surprised that even with this much larger data set, I was unable to get a much better match than my current best-fit. By my reckoning, there are only two worlds discovered so far that "match" Earth's temperature, pressure and gravity, as defined in my earlier thread; of those two, one is a tidally-locked co-orbital with a two-month-long day, the other has a day that's 2.6 days long. I'd judge that second world - Ovoky OW-B b5-2 A 1 - to be "Most Earth-like" in terms of environment, despite the extra-long day, lack of volcanism, two dim suns in the sky and smallish size. I will continue hunting for a better match for this, too.
I did an update news-worthy enough to post in the thread: I summed up all the sector overrides from the local Wregoe and Synuefe sectors, to better see how many ELWs and systems there are. You can find these here, on the "Wregoe sum" and "Synuefe sum" subsheets. As it turns out, roughly half of the systems in both sectors are "inside" the overridden sectors, so this gives a much better comparison than what was before.

However, especially when comparing to the Eol Prou sector, there is one notable flaw with the data: the total systems count does include systems that were uploaded prior to the introduction of the journal, so there's an unknown error there. Systems that do contain an ELW but aren't listed as such are likely contained there. No idea how many though.

Another possible source of error is that a good number of these overridden sectors reach across proc. gen. sector boundaries. I've decided to include them in both summaries, rather than exclude them from both.

The result looks quite interesting: although the Eol Prou sector is much closer (but not inside) to the core and the center, there isn't a big difference in the ELW ratio. The sector is not inside the core proper though, and as it stands, we have no sector inside the core that has more than 100,000 systems logged. Over that threshold though, there aren't big differences between sectors. This leads me to wonder if there really is any considerable difference between the core, the arms and elsewhere, at least in the ratio of ELWs. Unfortunately, there is no core sector that has more than 50,000 systems logged - on the database that I have, that is.

I'll try updating things if I can get a more recent dump of the EDDB body data.
I'll try updating things if I can get a more recent dump of the EDDB body data.
It looks like the dump stopped mid-October, and so far no sign of it returning :(
Good stuff, marx - thx!

As for the EDDB exports: Yes, they're unfortunately not being created anymore. I've been in touch with "Amiganer__Christian" in the EDCD discord who has been creating similar exports (don't ask me about the details/sources, though). We've gone through a few iterations to align his export with the structure/namings/conventions of the EDDB exports and it looks like we're nearly there. If there's interest I'll ask Amiganer__Christian if I can post a link to his export here (or the SQLite-DB I stuff that data into).
Sure, I'd definitely like that. I'd also prefer the sqlite database, since I'll be converting to that anyway, as before. (Thanks for the app again!)
This is excellent work!

As far as a replacement for the bodies dump, I'd love to see that, too. I've been playing with a way to estimate stellar metallicity from planetary data, and I've reached the "throw massive amounts of data into it" stage.
Thanks for reviving this thread. Just fooling around with the data in Excel, and found this interesting correlation, which supports the "multiple-path-to-ELW" theory. It also explains why it's so hard finding a "perfect match" for Earth.

It shows three clear "sets" of ELWs:
- the low-gravity, low-pressure set, below 0.87 G.
- the high-gravity, low-pressure set, above 1.03 G and around 0.75 atms pressure
- the medium-gravity, high-pressure set, 0.91 to 1.17 Earth masses and with pressures rising steeply from 1 to 4.25 atms.

Notice the big void in the graph around the 1.00-1.00 point and the general sparsity of anything in the 0.90-1.10 "box" around that point. Most of the ELWs around 1 Earth-gravity have much thicker atmospheres (2.5 to 4 atms).
Last edited:
Top Bottom