The "Steam" for this already exists: They are package managers.
Although most package managers also have package downloaders inside, the user can still install packages from files he downloaded/got provided these files are in a format understandable by the package manager.
Essentially, you can provide the package file and distribute it to the users, in whichever means you find useful, some of these being:
- CD with game and complete corresponding source.
- CD with game and *no* source at all.
- DVD with game and complete corresponding source.
- DVD with game and *no* source at all.
- USB with game and complete corresponding source.
- USB with game and *no* source at all.
- Public network service with game and complete corresponding source.
- Password-protected network service with game and complete corresponding source.
However note that in the case of password-protected content and with distribution not including the complete corresponding source, be sure to comply with the licenses that affect the project. This is very easy, just don't lock-down the user and be transparent and provide any *printed* document that the license tells you to provide to the user, even if the network service itself has a notice and even if the docuemnt is included inside the media. For more information, see this community-made guide to GPL compliance: [[https://copyleft.org/guide/monolithic/]].
Whicever you choose, once the package is provided, the user can use the package manager to install it.
On the subject of financial sustainability, this is an interesting challenge. One idea that might help is to think about software no longer as a "ready-made" product, but a constantly evolving one, which requires community contributions (financial (by means of paying for something), donations, testing, publishing/marketing, coding, art-making, license compliance enforcement, documentation, accounting, quality control, and so on), and considering the end-users as part of the community.
Be aware that donations and crowdfunding models which use network services to such ends generally have the following disadvantages *to the end-user*:
- No requirement for the end result, which *is delivered* to the end-user, to be free/libre software. And no requirement for the non-functional/practical data/works to be *at least* unlimitedly shareable.
- The network service being provided to contribute financially or by donations often requires the end-user (assumed to be non-tech person) to use non-free software automatically in the web browser (generally through JavaScript).
Also regardless of the network service choosen, donations and crowdfunding models are generally fragile and impossible to predict due to various factors.
However, although I'm not involved with it, Snowdrift.coop has various researches on the subject, mainly found in their wiki, also provides comparison on various crowdfunding platform service providers. However, I do agree with Snowdrift.coop in that Snowdrift.coop, once stablished and publicly open, will surely be the best of any of the providers ([[https://wiki.snowdrift.coop/market-research/other-crowdfunding]]).
Besides, Snowdrift.coop also compares crowdfunding platforms, and provides arguments for choosing crowdmatching over the rest (see previous reference); and also compares new forms of financial payment systems, and notes that GNU Taler might be the way out of this mess we are in ([[https://wiki.snowdrift.coop/market-research/payment-services]]).
Also, GNU Taler is payment system, not a new currency, it works with any currency, even dollars, Brazilian Real, BitCoin. More information in: [[https://taler.net/videos/taler2017eh.webm]] (talk by Sva, produced by c3voc.de, licensed under CC BY-ND 3.0). According to the talk, GNU Taler is currently needing at least one bank to implement it.