I agree. Instead of potentially forcing yourself into a corner to try to make a deadline and then delaying once again, it is better to just stay silent on any release window. As you said, once all the main features are on a live version, with only tweaks and other minor additions to include, then make an announcement. Seems to be the safest option for them and the community. Just my two cents.
Eh, it's really a lose-lose situation, because what people ultimately want is a great game (by whatever subjective standards any given person applies), right now. And some of them get really angry when they can't have what they want (just look at how worked up many, many people got over the inclusion-or-not of a single phrase in a dead language in another upcoming PDX release - this is just an illustrative example that should not be taken as a discussion prompt, because the topic got SO contentious that its general discussion is heavily restricted on this forum). Which is actually understandable, as games are part of culture, and culture is tied to identity and worldview, which are how we interpret and define our reality and ourselves, so challenges to our expectations literally feel like existential threats to some degree (even though they rarely are true threats).
For an example of a studio following your strategy, see Valve: mentioning Half-Life 2: Episode 3, then its reimagining as Half-Life 3, then nothing since then didn't stop the questions or speculation or screens-long rants on internet fora (or angry vlogging these days). Of course, now most people assume the project is canceled. They did release Team Fortress 2 as a complete surprise nine years after announcing it, and it was quite successful, so that approach can work in some sense, but it probably works best when your entire business has changed to being a publisher and storefront that happens to make a game every once in a while (or occasionally purchases entire studios that are already near completion of a game) rather than primarily a game developer, so your revenues (and investment capital) aren't tied to things like release quarter sales figures of your latest project.
The big problem that release dates try to solve (admittedly imperfectly) is that development is never actually finished. There are always additional features the team wants to make work, additional bugs to fix, additional optimizations that can be made. If you never have someone saying, "Enough, it's done, we're shipping this now," then development continues forever. In some cases, what you wind up with looks almost nothing like your initial software, and that can be a blessing or a huge problem (see Stellaris for example, which has radically changed almost every core mechanic from launch and progressively broken more and more systems and bogged down the engine with more and more state checks with each successive DLC/update to the point that it's practically unplayable from the mid-game on, while some game systems don't function at all, ever, or don't function properly from the start; even with that being the case, it has many fans who prefer the current state to the game that was originally released). The potential and contextual pressure for perpetual development is part of what has led to the enormously-problematic-on-the-balance software-as-a-service model (pressures of finance capitalism are the other part), where now nobody ever has a stable target for user platforms (whether it's a constantly changing OS that breaks more and more existing software with every update unless all of THOSE programs are continually updated as well, or a constantly changing game that breaks mods with each iteration) and users are constantly forced to adapt to new interfaces and (not infrequently) removed features.
PDX has tried to split the difference with their update/DLC-paywall model, and it can work well enough, though, as noted with Stellaris, that doesn't avoid all of the associated problems. There really is no one perfect solution - every development model has associated benefits and detriments. You're always facing constraints like those modeled by the project management triangle, and every decision has opportunity costs.
My personal adaptation to the new market norms has been to decide to never pre-order a game under any circumstances (not even EXCLUSIVE!!! content I think I might like - none of it ever stays exclusive, there's always eventually a complete edition or add-on pack, and I'm at the point where I'm not only not pre-ordering games, I'm waiting like five or six years after release until no more DLC is being produced and I can get a complete edition on sale; competitive multiplayer has some different considerations) and to never use software with a subscription payment model (including DLC "passes"). I still get to play most games I want to play eventually, and with the sheer number of games (even good games, in any particular genre) now available thanks to digital delivery all but eliminating publication barriers (and crowdfunding bypassing major publisher gatekeeping, though that only remains true as long as enough people DON'T adopt my never-pre-order approach), I'm never lacking something new to try out even with those constraints.
At any rate, I'll check out BL2 whenever it is eventually released and buy it if it looks like something I'll enjoy. I was an enormous fan of the first game (even the sewers on replay, which are not nearly as tedious as I remembered; I think my initial problem was that I was trying to be too conservative with resources in my first few playthroughs instead of charging in and slicing the abominations to bits), and was certainly exicted by the announcement of the sequel, though I haven't been following the updates exactly because I didn't expect to buy it immediately. Best wishes to the dev team and other fans; I hope you all weather this pandemic in good shape.