Hillary Clinton's loss to Donald Trump is a shocking upset that is likely to have disastrous consequences. A great deal is going to be written about what went wrong for the Democratic Party. But before we get too far into the recriminations, we should remember two crucial things. First, it's almost impossible to know how campaign tactics affected the outcome of an election. And, second, Clinton's campaign shows some of the potential perils of being too focused on the past.

Like most pundits, I have my theories about how the Clinton campaign might have screwed up. In retrospect, for example, it seems like the campaign made a mistake in making so much of its advertising negative attacks on Donald Trump's character. Given that Trump always had high personal negatives these attacks had diminishing returns, and Clinton missed an opportunity to highlight economic policy differences where public opinion favored her position. While it was not unreasonable to think Trump's particular unfitness for office created an opportunity to peel off suburban Republicans, it didn't work.

This is a plausible story, but to be frank it's just that: a story. Would Clinton using a more positive, policy-focused advertising campaign in the last month have allowed her to hold enough of the Rust Belt states that handed Trump the Electoral College? I have no idea, and there's no meaningful way to address the question.

Consider an example from the last election involving the popular vote winner failing to take office in January. For 16 years, I have been hearing people assert with the most sublime confidence that Al Gore's decision to distance himself from Bill Clinton cost him the 2000 election. There's no way of testing this theory directly, of course. But 2016 presented us with an indirect one. Hillary Clinton had a popular incumbent, one of the greatest political talents the Democratic Party has ever produced and without the scandal baggage and reputation for dishonesty that made deploying Bill Clinton a much more complicated question than Gore's critics will acknowledge, stumping hard for her. And, as a bonus, the incumbent's extremely popular and charismatic wife was also out on the campaign trail for the first major-party woman to be nominated for president. What was that worth?

Well, apparently, not much. Either the Obamas failed to move the needle, or they had an impact but it was swamped by other factors which can't be meaningfully measured. When it comes to campaign tactics, for the most part, nobody really knows anything. Be wary of assertions that there was One Magic Trick a candidate could have used to win an election, and be doubly wary when this magic bullet is an argument that the candidate advancing the policy ideas the pundit agrees with is also by remarkable coincidence always the best political strategy as well.

Admittedly, some arguments about campaign strategy are more concrete. The Huffington Post's Sam Stein has a smart postmortem arguing that the Clinton campaign deployed its resources poorly, neglecting Michigan and almost entirely ignoring Wisconsin, two of the traditionally blue states that Trump was able to flip. It's a solid critique as far as it goes, but it has two major flaws as an explanation for Trump's win. The first is that the Trump campaign was also not contesting Wisconsin, so Clinton making a play for it would have presumably caused the Trump campaign to respond. Maybe a fight over Wisconsin would have ended better for Clinton than the tough, high-spending war in Ohio did, but it's not safe to assume that it would. And, second, even if we assume that more money and attention would have allowed Clinton to hold Michigan and Wisconsin, it still wouldn't have been enough to win. Clinton still would have needed Pennsylvania or Florida or North Carolina, and she contested those states hard. It's possible that the same time and money with different messaging could have carried one or all of those states without a greater investment of resources, but such claims are essentially unknowable and unfalsifiable.

One rejoinder might be that while Michigan and Wisconsin ended up not being decisive, they could have been. Had Clinton carried Pennsylvania or Florida — both roughly within a point — then the decision to largely ignore Michigan and Wisconsin while investing in Ohio and Iowa, both of which Clinton lost by more than 8 points, would look really bad. It's a fair point. But the blunder the Clinton campaign made was to fight the last war, to be too slow to pick up on the particular threat that Trump posed in the Rust Belt.

This isn't to say that Democrats shouldn't analyze and try to learn from the defeat. But it's crucial to remember that the 2016 election is never going to be run again. We've learned for sure that Hillary Clinton should not be the Democratic nominee again, but I don't think that's something to worry about. Trump will presumably be on the ballot again, but as an incumbent with a record. What message and strategy the Democratic candidate should use will depend on who wins the nomination, what Trump's record looks like, and what the salient issues are. The 2020 election will be its own thing and should be treated as such. As Hillary Clinton now knows all too well, what we think we know about politics can be turned on its head very quickly.