Or, slightly less cynically, demand is always orders of magnitude higher for the first week or two after release, so it's only a short term investment in servers if they can be merged later.
In the future years when "study of video game history" becomes a thing, people will be able to examine and analyse old games like SimCity 2000 by actually playing them as they were. Since they have been archived and preserved by wide distribution.
Then as you get further into the future maybe you start to come across games where you have to say "sadly the only existing copy of the server code was lost in the great github fire of 2020, so all we can show you is this archived footage".
> so all we can show you is this archived footage.
You mean the tens of thousands of videos that will be posted. Historians will have no shortage of source material.
I'd actually argue that the videos are more valuable as they show first hand how the game was played at the time. Think a recording of a long-forgotten instrument compared to the actual instrument, which is more valuable?
I think there's quite a difference between playing and watching. Think of it like trying to understand Shakespeare based on clips from Baz Luhrmann's reinterpretation of Romeo and Juliet.
I absolutely agree that that's a loss, but that argument would seem to apply equally to all online games. But a game isn't less worth playing because it's online, right?
It potentially applies to any game without wide distribution of all required software. An online game where the server software was available to the public could potentially last forever, for example it is still possible to play early betas of counterstrike.
One solution may be to propose an escrow system.
A neutral non profit (or several) sets up an entity which exists to collect software binaries and/or source code from companies under strict NDA.
At some event, either after a period of time or at the time in which the service based on that software ceases to operate; the software is released to the public (either for free or for some payment to the escrow entity).
In return the company gets a "no abandonware" sticker it can use in it's marketing/branding which gives consumers confidence in their purchase.
I seem to recall that the reason for the cloud based server was to leverage more number crunching power to yield better AI. Running it locally might not help. (Though, it certainly is better than no server, and it'd be interesting if they let people tune the fidelity of the processing so that it could run locally, or on a private server.)
I highly doubt they are providing server resources for every player that are more powerful than the player's desktop or laptop computer. Surely that would be extremely expensive.
I agree - at a very low 100MHz per player that's 25 players to a 2.5GHz Xeon core, 100 per CPU, 200 per server. Assume 100,000 players, that's 500 decently spec'd servers plus redundancy, bandwidth, storage, sysadmins, etc.
Your probably right, but maybe if you are doing all the calculations for everyone in one place you can start to make large scale use of dynamic programming techniques over time?
