Warning: wall of text post. If not your thing, scroll down to the next one.
As someone that has been participating in versions of this discussion for more than a decade now, it has been very interesting to watch how it has evolved over time. At the heart of all sports arguments is the idea that we can guess-timate what one player would do in all situations based upon what they do in their current situation. Thus, in most debates one side will establish their set of arguments for what one player could or couldn't do if the situation was swapped. With Garnett, though, he keeps over-coming obstacles that he wasn't supposed to be able to, so those that argue against him have had to move the goal-posts, if you will, changing their reasoning for why he isn't as good as he is when their old theories went up in flames. Let's take a quick walk through history:
1999: The real "start" of the KG vs Duncan theme. In 1998 Duncan had burst onto the scene and won RoY but Garnett had signed the biggest contract in NBA history that many blamed for the lockout. At that point Duncan was the better NBA player (more polished coming out of college, posting better stats on a better team) but everyone could see that KG was growing into his potential and that #21 vs #21 was shaping into an interesting rivalry. The initial argument against KG was one of "substance" (Duncan) vs "flash" (Garnett), that Duncan was quietly productive whereas KG was all potential but couldn't do it in the box scores so he couldn't really compare. Then, in the '99 playoffs the Spurs played the Wolves in the first round. KG outplayed Duncan head-to-head statistically, even giving that rampaging Spurs team the only loss they would have during the whole Western Conference playoffs. But Duncan and Robinson's Spurs won the series, on their way to an easy title. And the new meme was established that Duncan was a winner while KG was empty stats.
Fast-forward to 2003. By now the "winner gene" argument has taken firm hold, because there's really no other explanation for why Duncan is better. KG has caught him statistically, started pulling 50-win teams out of little support, and has finished top-2 in the MVP vote twice in the past 4 years. But Duncan has 2 MVPs and 2 titles in that stretch, though. So the argument is that KG is great, but he just doesn't have Duncan's "winner" gene. Then, the Wolves traded trash for fading vets Cassell and Sprewell. Interestingly, at the time, NOBODY thought this was that significant. You'd hear folks say it was the best cast Garnett ever had, but it wasn't really expected to result in much because a) KG didn't have the "winner" gene, b) Cassell and Sprewell, both perimeter players in their mid-30s that had recently been considered mal-contents on poor teams, weren't really expected to have much left, and c) the rest of the Wolves were still made up of cast-offs (Hassell, Hoiberg, Erv Johnson, Olowokandi, Madsen, Hudson, etc.) and Wally was injured. But then, Garnett turned in one of the more dominant MVP seasons in history (according to the voting), led the Wolves to the best record in the West, and was a healthy PG away from possibly competing for a title. The "empty stats" and "not a winner" arguments were proven invalid, so the goal-posts had to be moved again.
Over the next few years, the new meme became that Duncan was more "clutch" than Garnett, a better defender than Garnett, and his interior offensive game was more conducive to winning than Garnett's. These arguments were helped because the Wolves started missing playoffs, while the Spurs were winning 3 titles in 5 years. People don't like "excuses" like, oh, that the Wolves were often horrible outside of Garnett to the point that they scared small children. No, the narrative required that Duncan had greater success because he was a greater player. And since Garnett was still out-doing Duncan regularly in the box score stats, there had to be some "intangibles" (like clutchness or post-game impact on teammates) or at least "not-measureds" like defensive impact that would have proven that Duncan had really been better individually than Garnett all along.
The problem was...actually there were 2. Problem 1) was that right in this period we saw a rise in the study of "advanced" NBA stats, which looked at, among other things, how a player performed in crunch-time situations and how that player's presence on the court influenced the play of the team. And those stats, unanimously, said that Garnett was either better than or similar to Duncan. His clutch performance was just as good...his impact on the team offense was better...and his impact on team defense was also as good or better. This flew directly in the face of the anti-KG argument. And then, problem 2) happened: Garnett went to Boston.
The Boston years. See, we forget it now, but very few expected the '08 Celtics to actually WIN the title. People thought they'd be good, but not true contenders. At least in large part because they didn't think that Garnett was "that guy" anymore. People questioned the lack of depth on the team. Lots of folks thought that the defense wouldn't be good enough. So that's why, when KG erupted and the team went darn-near undefeated through their first 32 games, folks were shocked and KG was getting massive MVP buzz. He ended up finishing 3rd and winning DPoY, and also leading the Celtics pretty clearly through the playoffs to the title. But Pierce won Finals MVP, so this established the new line-of-attack on KG that still extends through today: "OK, he can win, but only if his team is stacked and he's not a real #1 option". Unfortunately, the next season Garnett severely injured his knee and wasn't healthy enough to change the answer to that question in '09 and '10. He seemed to regain his health in '11, but by the end of the Heat series he seemed worn down in a way that one has to fear is age-related. Now, at age 35, there is legitimate question as to whether KG can still bring it at "that" level for another champion. If he can't, 2008 will stand as the only evidence one way or the other of what he could do with a championship-caliber cast.
So, look where we've come from. In '99 KG was just flash, no substance in the box scores. In '03 he was empty box score stats that couldn't win. By '07 he was better statistically than Duncan, but he lacked his intangibles and defensive impact. Then, he proved that he could lead a team to a title then got injured before they could defend it, and the new anti-KG bench-mark is that he can win, but he must really be a second option. Oh yeah, and the "advanced" stats pretty unanimously peg Garnett as the best, most impactful player of this generation...but instead of making people re-think their opinion on KG, what's happened is that it's caused a large degree of skepticism in those stats and slowed their acceptance by the main stream. Quite literally, if the roles were reversed and the advanced stats said that Kobe was the best player of this generation with Garnett hovering in the top-5 I guarantee that advanced stats would be pushed more publicly and would be a more accepted part of the way the NBA is covered.
And likely, this is where the KG legacy question will eventually be argued. I still think these Celtics have at least one more championship run in them, and if that happens there's still time for the story to change, but for the most part we've seen the lion-share of Garnett's career. I think it's been patently obvious that he did as much with what he had to work with as any player of his generation could have, including Duncan. And I've got this still-developing body of advanced stats that support what I believe I saw. But no matter what, there's no time machine. We can't transport back to '99, give KG a cast as strong as Duncan's, and really see what would have happened. All we have to work with is what actually happened. And what happened is that Duncan had the better team, had better team success, and won more individual accolades (which are strongly correlated with team success) than Garnett in their primes. On the other hand, KG had better individual box score numbers in their primes, better advanced box score marks, and dominated in the +/- stats while adding some pretty strong accolades and accomplishments of his own.
In the end, there is ammunition for both sides to argue their cases and the general NBA follower of today definitely leans toward Duncan. I still think Garnett was and is slightly better. But in the end, both are historic players that should absolutely be at the top of any ranking of best PFs in history. Names like Malone, Barkley, Pettit, Dirk, Mchale can all be discussed as well. But for mine, the 2 best PFs that ever lived both wore #21 in the Midwest division for much of the 2000s, and the greatest one ever currently wears #5 for the Celtics.