NAM Team policy has always been strongly against the continued distribution of old versions, save for very few special cases (translations, and the debacle that was NAM 31 AKA "NAM Vista"). While some individual team members might still some them lying around, they're effectively treated just as unreleased content is--not for public consumption--and there is indeed no official internal team archive of past versions.
To be blunt, keeping in mind that I've been a part of 16 NAM releases prior to NAM 38 (not counting those hideous "point releases" of NAM 31.x ignominy):
Old NAM versions are rubbish. And if they aren't rubbish when compared to the current release, then we're doing something wrong.
This of course, begs the question: what makes old NAM versions "rubbish"?
Technical SupportThe issue of technical support is the main one we've cited over the years as the reason for our embargo on old NAM versions. We're an all-volunteer team dedicated to producing a freeware product--and for the last five years, we've been a pretty small team at that. The reason we do this is because we enjoy expanding and improving the array of transportation network content in SimCity 4. Technical support is a responsibility that comes with the territory, but our goal is to minimize our need to provide it to the greatest extent possible, such that we can focus on the labor of love that is developing new versions. To that end, we've strictly limited those responsibilities to the most recent version, and go to great lengths with our testing routines (which have really stepped up this year) to ensure that it is a solid product, fixing many bugs and glitches that existed in previous versions in the process.
While, for instance, Microsoft might keep support for old versions of Windows going for a few years, in those cases, (a) the user of the product paid for it, (b) upgrades to the newest major version usually come at a cost, and (c) being a large company, they have paid support staff. Even with larger open-source projects that might offer "Long-Term Support" (LTS) for old versions, they have a larger staff that can devote some resources to support, and some of those involved in development and/or support might even be getting paid for their involvement. None of this is true of the NAM or the NAM Team.
Some have asked, "can't you just allow the old versions to still be available, but with a note that they're no longer supported?" The answer, from extensive experience in the SC4 community, is a rock-hard "no". I'll point to Exhibit A: even though we stopped distributing and supporting NAM 36 back in July 2020, with the release of NAM 37,
that hasn't stopped support requests for NAM 36.
Additionally, back in the early days of my tenure on the NAM Team, when NAM 21 was launched in June 2007, one major site ended up being slow on the uptake, due to their file handling requiring anything larger than 10MB to be hand-uploaded by the webmaster. The result: a rash of users mixing and matching bits of NAM 20 with things that required NAM 21, and wondering why their menu buttons didn't work and gave them red arrows. Dealing with "red arrow" tech support was basically a full-time job until that site updated their NAM download to NAM 21. And it remained a nuisance until we went back to the "Monolithic" approach with the notorious NAM 31.
Server/Disk ResourcesRecent NAM versions, since the switch back to the "Monolithic" approach, have run in excess of 300MB for the download, and are presently over 700MB. (The big jump with NAM 37 and above has to do with changes in compression with the new installer.) While the NAM "proper" in the so-called "Modular" releases from NAM 21-30 might have been quite a bit smaller, that's deceptive, considering that the full set of features that are now included in the mod itself were separate downloads, which, once combined, would effectively add up to close to that amount.
The NAM Team has, for almost 8 years, used a shared Dropbox folder as the primary means for sharing in-development files. It's on a free account with a 2GB limit. The amount of content in there--including a fair bit of unreleased stuff--runs about 1GB without any fully-compiled NAM builds on there. We have room for the most recent developmental build that's in testing, and that's it. It's largely proved to be a reliable system, and it doesn't cost us a cent.
Ease of Use/Legacy ContentAs
jeffryfisher aptly noted, we're often trying to upgrade the general user experience with the NAM with each successive release as well. The new installer we introduced, while having some entry barriers (Java for the Windows users) has eliminated the "installer forgetting files" issues that plagued NAM 36 (and some releases prior to it), and has vastly simplified the process of providing tech support to the small-but-very ardent faction of NAM users running the game on macOS.
The move away from the old-style puzzle pieces to draggable items and FLEX pieces (individual FLEX pieces are able to handle functions that would require upwards of a dozen old-style puzzle pieces) is also allowing us to provide users with shorter TAB Loops (and eventually, fewer and more streamlined menu buttons). This means that the user is getting more functionality, but with fewer menu items to navigate, subverting the notion that expansion means more complexity. I'll note, we haven't added a single old-style puzzle piece to the NAM in over 5 years.
Old NAM releases aren't necessarily simpler, and indeed, if one goes back to the pre-Monolithic releases, the process involves running multiple installers, and making sure you're using the correct versions of various plugins with the correct NAM version in order to get everything up and running. Go much farther back than that, and many basic features that NAM users take for granted today--like Roundabouts and Diagonal Streets--didn't even exist.
There may be some users who have complained about features "going away" in successive releases, but this is largely a myth. We go to great lengths to ensure that content built with previous versions will work the same or better with the latest version, only breaking that continuity in cases where providing legacy support is simply no longer feasible, or when old features are found in further investigation to be outright broken.
For example, in the late-00s, the old "10x commute" and similar options for the traffic simulator were removed from the NAM, because they were found to not work as advertised. We also killed the Auto Road Turn Lanes with NAM 31, after finding that their implementation would effectively prevent a lot of exciting new features--like Draggable Fractional Angle Roads, Draggable Road Viaducts, and the Network Widening Mod--from being able to operate in a reliable and stable manner.
With NAM 37 and beyond, we have pulled out support for the base Maxis Rail, and for some cosmetic reskin mods that had some semblance of a userbase (the Bullet Train Mod and the Alternate El-Rail), and haven't yet offered up a suitable replacement for the users who are not on board with the RRW and/or stock Monorail and El-Rail. This is in part due to our needing to prioritize the solution to the existential threat to the NAM Team--runaway development cycles, like what we had with NAM 37. Now that we've solved that, we're able to return to that issue.
Fair PointsThe fair points that can be brought up are (a) the number of options in the installer (which may be complex to unselect properly to get a more basic installation), and (b) the technical requirements (the larger default NAM Controller requiring the 4GB Patch and a 64-bit OS, requiring users with less RAM and/or on 32-bit OSes to manually compile a Controller themselves).
However, treating old NAM versions as a solution for these users with old NAM versions is foisting upon them all the bugs and inefficiencies of that
rubbish. If this userbase is indeed on the newbie/novice end of the experience scale, as speculated, they're going to want tech support, and they're not going to get it with old NAM versions.
The idea I've floated as a solution to this is to produce a "NAM Lite"--effectively, a stripped-down, basic version of the most recent NAM, which would be eligible for tech support, since it's operating on the same "core" as the full-blast NAM. The matter of just what would go into "NAM Lite" is still a question, and we're open to feedback as to just what that package would entail.
-Alex