Can the cost of running the game infrastructure really be significant if, after 4 years, Frontier are still prepared to keep the Legacy infrastructure running for an ever decreasing player base?
and
Yes and no. If what @Sylow wrote about the stripped down infrastructure, the costs for the servers are minimal (but not zero). But maintenance is not only server costs, it is also personal costs and those could handle legacy and live at the same time. Do you got any numbers, how many players are active on legacy or live? Mind that they are still migrating people away from console and legacy...
On personal costs and so on: that is very much true. And well all "know" that there are over a hundred people working on ED. And yes, this is a sarcastic statement. As they officially also were working on ED on those years, where nothing happened, merely bugfixes. And after a year, we got half a page of bugfixes, most of them insignificant.
With the sarcasm out of the way: i never was in this part of business and i am no insider for FD. So i can't say anything about their personal costs. I can only guess, and will come back to that below. I first would like to look at the infrastructure part, because that's where i have professional knowledge. (Or rather had. My work till 2008 included server administration, including planing resources needed for to keep the systems operational. )
As the info on their infrastructure is from around when the game launched, i have to make some guesses. The first one is, that they still rent virtualized servers from Amazon. It could also be that they switched to another provider, host things elsewhere, etc. An interesting statistics would be
this. I tried to get equal information directly from Amazon. But while they are ready to provide a pile of information (for example see
here), i found it not to go back as far as i would have liked. So i went with the first link i put there. I don't know the hidekazu-konishi.com, but the data at least looks consistent and plausible enough to me for the purposes here.
So we know that if they are on the same servers, they should actually be paying significantly less per year and GB now than at launch. On the other hand, in pure theory having two instances of the BGS running in parallel can double the GBs used. In practice, the numbers should be way lower than that, though. As values decay and fall out of the database when getting close enough to the seeds default value, the space used mostly is inflated by explorers and colonisation.
Explanation for why i conclude that: everything in the "default bubble" before Horizons returns to default values in the run of time. (About everything in the world is default in the bubble. Stars, planetary bodies, Stations, their names, trade goods, their prices, which factions are there, their influence, etc. If no player acts on it, the game over a several weeks gradually moves everything back towards the default value and once that is reached, it will be deleted from the database. After all, the data can always be restored from the seed. )
All the deviations are created by players. A low number of players also create a low number of deviations. In pure theory, if a dozen players operate in the same system, the game still only has to store the data for one system. In practice a low number of players in the bubble can create a comparatively high load. As their overlap in systems visited will be limited. Yet the total impact of a low number of players is low.
Now why i differ on exploration and colonisation: the info for first discovered, first scanned and first footfall are not default values. They are created by the players moving around and have to be stored on the servers permanently. So this is permanent growth, without a reduction system behind it. The same is true for colonisation, players fill systems with new data, based on their decissions. But both still deserve a little bit of a closer look.
First exploration: yes, there are many systems out there. And the database size for that can only go up, never down. (At least according to how the game should work. ) But the growth rate is both limited by the number of active explorers on this instance of the galaxy and by how little information is actually saved. On any normalized database, which is the minimum i expect, the maximum per planetary body are three values saved. (Discovered, scanned and landfall. According to what i tend to see when looking at systems. If there's more by now, then let me know. I am not an active explorer. ) And the values are not the space wasting strings, but merely the player ID. So probably two to four bytes per entry, depending on for how many players they planed for. From my limited experience with exploration, while i found systems with almost 50 bodies, there's also a lot out there with only the main star or merely one or two additional bodies. So for simplicities sake, i now claim that the average might be around 10 bodies per system. Based on that really rough guess, merely the text i wrote here up to now can store the information for 150 systems explored. (Now look at how much text the forum here absorbs each day. People will have to do a LOT of exploration to keep up with that. )
For Colonisation i have to guess much more. There's a pile of data for stations (space and ground), trade goods, prices, etc. to be saved, which otherwise the world seed covers. I would guess that the game can still use default values for trade goods and their prices based on the installed economy. So a part of what the old presentation showed might still work here, but i only guess that, i don't know how they implemented this. And while there is a kind of clean up system called upkeep, it is slow in cleaning up. Thus i would guess that Colonisation lead to a noticeable growth of the server space use.
But now circling back to above: Colonisation is only on Odyssey, it does not affect the legacy infrastructure. And while exploration can only add to the space needed, prices for the servers seem to have dropped, as storage space prices have fallen over the years. (Again: i have no insight, on which deal FD has with Amazon. I can only work with the numbers for which Amazon offers their services to normal customers. )
Guessing that the lauch of Odyssed did not significally increase the number of explorers on the legacy servers, i thus dare to conclude that most likely the costs per player, even for explorers, have dropped significantly for the legacy servers. I could now pull some numbers out of a low placed orifice, but that's not helping anybody. Having two instances of the galaxy running creates some overhead. But the falling server prices over the last decade should more than have compensated for that. Only for the life servers Colonisations might have countered it up to some degree. (Again: all guessing. I have no idea, how Colonisation is implemented, how high the impact on space used is, etc. )
And after all this, i want to circle back to personal costs. Because: the servers are rented out to Amazon. So all the costs for server maintenance, hardware upgrades, software updates and people watching over it are paid with that. (Sidenote: I would expect that Amazon is not responsible for rolling out new version of the database and fix issues there. Neither do they do customer support. These things are on FDs side. But the first things are basically part of the games ongoing development, the second is related to the number of actually active players. So they have to be seen within the scope of the complete picture. ) Unless there is real crazy stuff in the code which we are not aware of, i can not paint a scenario bad enough that the inflation of space can somehow keep up with the price development from Amazon. So unless the space used (which is why i focused so much on it before) has skyrocketed insanely, the costs on that side should actually have noticeably dropped for FD over the last decade.
With all of that down, i think i just also answered why ArenaNet can easily still keep Guild Wars 1 running for example. It probably is merely an additional instance on their infrastructure. (As far as i know, they operate their infrastructure themselves. ) Being a virtual server means it creates basically no additional costs for maintenance, it merely can take a small amount of processing time, memory and storage space. Directly proportional to the number of people playing at that time. And generally over the last decade prices for all of them have dropped a lot. (With the exception of memory now, thanks to the AI craze, being crazily expensive. But that's another story. )
With all of that written, i have to point out that i can not tell anything about the costs of creating new content. I very much doubt that we indeed still have the 100 people working on ED, which we supposedly had for many, many years. If they'd still be there, then i would wonder what they would be doing these days, comparing how much content a year we got in old time to how much we get these days.
So, we don't know how many people still work on ED. We don't know how many people actually buy the new ships and stuff, how many buy which packs, paintjobs, etc. All we know is that FD in their financial reports to the public state that FD is lucrative. I do not have the exact numbers at hand, how lucrative it is. And if i had the numbers, i still would have a limited idea on how to read those finance-guys-numbers. I remember that some time ago, i read (either here or on reddit) some posting, where somebody tore into the more serious numbers. And his conclusion was, that ED is not just scraping by, but actually doing quite well. But behind "somebody did this" and "i remember", this is nothing i would bet even one cent on. So i have to stick with the the interim and annual reports, which i linked a page or so earlier, where FD at least claims that ED is lucrative again since almost two years.
And after all this text, i return to my agenda: If the game is lucrative, then the way they treat us here is not out of desperation. It's intentioal bad faith behaviour. And i really wish FD would finally understand and fix this. But how to actually make them even register this notion, i have no idea.
PS: By now the text is long enough to cover about 325 systems fully explored. I want to see the explorer who can explore that many systems in the same time it took me to write this.
