Every year at this time the Kansas City Federal Reserve Bank hosts a conference in Jackson Hole, Wyo., a major throw-down where economists from outside and inside the Fed get together to talk monetary policy. For a few days, the mountain air ripples with dense discussions of interest rates, inflation dynamics, unemployment, risk, recession and so on.

As is typically the case with Fed events, everyone’s listening for hints about the central bank’s next interest rate move, and this year, Chair Janet Yellen’s big talk on Friday will be even more closely scrutinized than usual. That’s because our understanding of the Fed’s thinking is uniquely fuzzy right now. As one wag put it the other day, “Fed types do not just contradict each other, now they contradict themselves.” This is a reference to recent, very insightful work by the president of the San Francisco Fed, John Williams, whose inner dove (keep rates low) is fighting with his inner hawk (raise rates soon).

My purpose in this column is to explain what that bird fight is about and why this year’s meeting is potentially significant. One way to think about this is to consider three numbers: 5, 3 and 2. These (roughly) represent the Fed’s current long-run targets for the unemployment rate, the interest rate and the inflation rate.

Each of these numbers is probably wrong, and at least some of the officials know it.

According to their published forecasts, Fed officials think the lowest unemployment rate consistent with stable inflation is just below 5 percent. Their long-term neutral interest rate is 3 percent. For them, stable inflation means price growth of 2 percent; that’s their inflation target (why 2 and not 3 or 4 or 1? See here.)

A number of problems/anomalies are quickly apparent. First, note the figure below. It shows that a) the Fed has consistently been undershooting its 2 percent inflation target, and b) as actual unemployment has hit the Fed’s full employment rate, inflation hasn’t sped up at all (nor did inflation fall much when unemployment peaked). The model predicts that as unemployment falls to its lowest sustainable level, inflation should tick up. But it doesn’t. This may well be because, as I’ve shown elsewhere, the estimate of full employment is too high. After all, it’s an estimate of a concept, not an observable number, and the margin of error around that estimate right now is historically huge.

The next problem is that Fed officials think the neutral Fed funds rate — the rate that’s neither stimulative nor contractionary — is about 3 percent in nominal terms. But for interesting and important reasons, including, according to Williams, “shifting demographics, slower trend productivity and economic growth, emerging markets seeking large reserves of safe assets…interest rates are going to stay lower than we’ve come to expect in the past.”

Economist Larry Summers has been hammering this point for a while, and he correctly underscores its connection to another problem, a more nuanced but even more serious one: such persistently low inflation means that the Fed is increasingly exposed to the problem of the zero lower bound.

In downturns, the Fed wants to reduce the real interest rate (the nominal rate set by the Fed minus inflation) so as to stimulate more investment spending. But generally speaking, it can’t lower the nominal rate below zero (yes, some central banks are pushing rates below zero, meaning member banks have to pay the central bank for holding their money; weird, right?). Higher inflation, on the other hand, can mitigate this problem by lowering the real rate. That’s why there’s a strong argument that the Fed’s 2 percent inflation target is too low; a higher target, which would encourage higher inflation, would give the Fed more room to maneuver when the next recession eventually hits.

So, to recap, 5 and 3 (the full employment rate and the neutral funds rate) should be lower and 2 (the inflation target) should be higher.

Like I said, dense stuff, but it’s all actually pretty clear to numerous economists. Still, Fed officials, including Williams, are going on about getting behind the inflation curve, the need to raise rates, and making cryptic distinctions between “letting up on the gas” and “tapping the brakes.”

Markets are confused, and showing it on a daily basis (see this chart on the dollar’s reaction to the release of the Fed’s July minutes last week).

Williams does, however, make a provocative point about the 5, 3, 2 problem and this current state of confusion. He seems to be saying that the Fed can’t just make an immediate, unilateral decision to start targeting different numbers or, more radically, start targeting totally different variables (like the level or growth of nominal GDP, or better yet, wage growth). Instead, he argues implicitly that while the Fed thinks this through and perhaps gets markets, the public and the politicians ready for some big changes, policy incoherence is the price we’ll have to pay for now to avoid any sudden market disruption.

It’s a reasonable take, but I’m not sure it’s right. I suspect it would be a relief to many close Fed/econ/market watchers if they made a clean break with the current, seemingly inapplicable model. And given what we know about that model, it would be a huge and consequential mistake to preemptively slow the current economy down just as wage and job gains were beginning to reach middle- and low-wage workers.

I’ll be listening carefully, amidst the moose and elk calls, for signals from the mountains of Wyoming that the 5, 3, 2 problem is understood and that plans are being made to adapt the policy to this reality.