If I had the resources of D&D Beyond, I could have done it myself a few times over by now.
If the D&D Beyond team had the resources at the beginning that they do now, and — just as importantly — the support of the people with the money, I'm sure they could have done a much better job. I think it's a tad too easy for a lot of us to talk about how we could have done a much better job than they did given their much better financial situation now as well as the benefit of hindsight. You mention that you
tried to create your own character builder more than once and acknowledge the difficulty that lies in such a project — imagine that's what you're trying to do with a team of 2-3 developers and no real financial backing in the hopes of getting something out the door that people would be willing to actually pay money for with the ability to scale as an open, web-based application would need to. Like you said, not even remotely easy to accomplish. Possible, yes? But with a lot of support and on the backs of people who have already made mistakes.
It is not easy to implement, but the initial design choices really do make a huge difference, as someone pointed out.
Absolutely. In an ideal development lifecycle, most of the engineering effort should be in the design and architecture phase. The implementation
should be fairly straightforward because you've already done all the heavy lifting... and, of course, we all know that we can see into the future and anticipate every single possible problem and never release bugs because our code is always perfect

.
By all accounts, the development lifecycle of that first iteration of D&D Beyond was far from ideal, and they haven't really ever gone in and done any significant refactoring (although they seem to be on the cusp of releasing major changes to at least one aspect of their platform) that might warrant a new major version... they've essentially been building on top of the original foundation as best they could, because the people who gave them lots of money had a very specific vision for what they wanted the platform to be.
you can certainly create software that would handle 90+% of homebrew, and WotC ain't doing nothing in UA that most of us here haven't tried in a homebrew context at one time or other.
You can, if that was something that was even on your roadmap to think about in your architecture, and despite what some people seem to think, homebrew
classes are not things that the majority of the player base these days really use.
Most 5E players have never played another version of D&D before. The market
exploded a few years ago, thanks to things like
Critical Role and
Stranger Things, and many players don't see much beyond just the core rulebooks — a huge swath of the market is casual players who aren't buying up every single new book but rather just playing for free in some local game store or at the library with their classmates. Homebrew in general isn't necessarily a niche market, but most players have never heard of Matt Colville or his Illrigger (which first saw light of day only a couple of years ago, long after D&D Beyond came out) or the Pugilist on DM's Guild, arguably the single most popular homebrew class that wasn't done by someone with the first name Matt.
As I mentioned in an earlier comment in this thread, I've actually had in-person conversations with some of the development team in the past who indicated that homebrew classes were simply not a priority for the business because there wasn't actually demand from that many users for it. I'm sure demand has increased for such a feature since the team more recently mentioned that they potentially have that on their roadmap, but it's far future stuff because it's not something nearly as desirable as something like a VTT... and you can be sure that Curse as a business is going to push the team to make the things that the majority of their user base actually want.
And, as some in the thread have also pointed out, "cheating" (by making each subclass a set of subclasses, one for each class it was allowed for, which I'm sure would have been perfectly satisfactory for the majority of users) would have made it even easier.
Sure, but that's still time that the development team would have to put into doing it, when they couldn't even be reasonably sure that Wizards would actually follow through with it and when Curse obviously has much bigger things they want the team to be working on. As someone who has been in the industry for as long as you have, you should know just as I do that the engineers are rarely the people making these kinds of decisions. There are a lot of things about the platform that my team and I work on that I would love to refactor, optimize, enhance, or otherwise improve, but there's no way my company lets us invest the necessary time to do everything when there are tons of features that paying customers actually
want us to do.
To be fair, the guys at Shard have a lot of hindsight to go on, and they were able to learn from DDB's mistakes.
This, 1000%.
Any class that wasn't created by WotC is technically a homebrew class, and, other than a privileged few (e.g. blood hunter)
You mean privileged
one. LOL
There have been a few unofficial subclasses introduced into D&D Beyond although the only ones that stuck around were ones by, you guessed it, Matt Mercer.
They had the Runeterra stuff up for a little while, but that was taken down a long time ago.
Suppose I want to play a feywalker from EN World's
Masterclass Codex (which I am in fact doing in one of my games). Suppose I want to play an
illrigger from MCDM. These are very high-selling products, but a character creator that doesn't allow homebrew classes can't handle them. That seems problematic if DDB wants to "win" the market in the long run.
Again, not really, since the majority of players don't use homebrew classes in the first place. While I don't deny that products like those you've mentioned make decent coin, the numbers of people who are actually buying these things up is a miniscule fraction of the total 5E player base. If Wizards is to be believed, there are over 50 million 5E players right now. The Pugilist class, arguably the most popular class on sale on DM's Guild and objectively one of the most highly regarded homebrew classes created for this edition, is an Adamantine seller, which means that 5,001+ units have been sold. Of course, we don't know what that number is, but I highly doubt it's anywhere even remotely close to even just
one million. Heck, I'd be surprised if it's even 100,000.
As far as whether not having this feature will prevent them from "winning" the market in the long run, what I can say with a lot of confidence is that Curse is going to prioritize features that users have actually indicated that they wanted since they want to, you know, make money. If everyone and their mother wanted a homebrew class feature for the platform, they would be prioritizing that. They aren't.