(To Sopko, specifically, before I get into my general argument: if booting exists (and it does), then if you play a game that is on the cusp of being booted, you really have little to complain about if the game is booted as or shortly after you play it; in fact, you're kinda asking for it. If a game is probelmatic enough to be booted, chances are one person playing it isn't going to help much.)
Okay, look. It's all well and good to say "shrinking the roster makes no sense", but hold up for a moment. Let's take that idea and follow it to its logical conclusion. If the allure of the DL is indeed how many games are represented, then why wouldn't we just rank as much as possible? After all, by saying no to booting you are also saying there is no such thing as a bad rank. We could do this.
If we wanted to destroy the DL.
Exhibit A: Best of the Not Ranked. I will be quite frank about my opinion here: NR sucked and I don't miss it at all. Why? Because it's simply NOT FUN to barely be able to vote on things. Unless you've played 300 RPGs like Niu, then a pool that broad is, due to the fact that even a single dueller you can't vote on disqualifies your voting on a match, going to create severe voting issues. Note that the effect is quadratic; doubling the number of duellers you can't vote on (making some simplifying assumptions) creates four times as many matches you can't vote on, and that doubling happens pretty fast as you rank more and more obscure RPGs.
And all that non-voting is very unhealthy for a social site. Go back, and look at some Not Ranked season topics. Notice anything missing? Why yes, you may notice a distinct lack of discussion on the vast majority of matches, certainly far less than the active DL season topics. When far fewer people can vote on a match, there's much less potential for discussion of the match, and discussion of a match is the site's reason to exist. Without that, it descends into little more than a popularity contest.
Okay, so, ending the tangent of "ranking everything is extremely stupid", we get to booting. How do these connect? Pretty simple. Having accepted we are not ranking everything, it follows that we must have standards for what is rankable. While there are several standards for ranking (plot relevance, dueller memorability), it's hard to argue with "People can vote on this" as one of the single biggest criteria. I think we can all accept this, right? This ties into my earlier comments: creating more matches that few people can vote on is unhealthy for the site, and you do this by ranking poorly-played games.
Okay, so, how do we know if people can vote on something before ranking? The answer: we don't. We can guess through a variety of means, and those guesses are pretty accurate, usually, but it's impossible to know for sure until we start seeing some matches in practice and looking at, yes, PERCENTAGES.
Booting, then, is where we behave like adults and admit that, yes, we made a mistake. We shouldn't have ranked ____. Correcting that mistake is healthy for the DL. The sheer ABILITY to make mistakes is healthy for the DL. It means that we can experiment by ranking shaky ideas and see how they turn out. WA4, somewhat unbelievably in hindsight, got some opposition when it was up for ranking, as people pointed to its sales numbers and suggested it would fail. As we know, it was a great success; the knowledge that booting existed had it failed miserably was a safeguard that helped it get ranked. On the other hand, Baldur's Gate 2 was another very reasonable experiment - people honestly believed that there might be many casuals who had played the respectably-selling game, or that we'd attract more PCRPG fans to the DL. Neither happened, the game flopped, so we could boot it without permanently damaging the DL.
I look at the current ranking topic. I see Mana Khemia, I see Wild Arms 5, I see MMXCM. There's no Dragon Quest 8 in the bunch. Any of these games could end up failing miserably. I don't consider the chance negligible for any one of them (the odds of extreme failure are slim, granted). If we couldn't boot, I would start no-voting all of them, and many other new ranking ideas. I assure you I would not be alone. This would be bad for the DL.
Booting has a further purpose. That is to account for shifts in the DL votership and general opinion. When we first ranked things back in Contenders days, Nall and Ruby were good enough to be ranked; we booted them as opinion shifted to the non-consideration of clearly defined battle forms. General Leo was ranked and even enjoyed some success, but 30 seasons later his battle form was considered too borderline plot-ridden and unscalable to be suited to the DL. And finally, games like Arc the Lad 4 may have enjoyed adequate votership in the early days of the DL, but as more and more new voters came along who hadn't played it, it started to struggle more and more to meet the criteria of rankability. The world is not static, so a booting mechanic is needed to address changes.
Whether you want to boot many characters or few or even none at the moment, the mechanic of booting is one that is very necessary for the DL to continue functioning. This is true whether you see it as a business or as a social site (and I agree that you should see it as the latter). It allows criteria of rankability to be upheld even after a game is ranked, to correct mistakes, and to adapt to changing times.
