There's no problem mapping a shoreline. Until, that is, you look at it too closely. Suddenly the definition of this clear line loses resolution. Even if you decide to draw your map at a particular time of day and thus plan on drawing a gray zone to account for the tides, as you drop your scale to the size of a grain of sand, indeterminacy overtakes accuracy. Do you count this grain or not? It's half under water, half above water. And now, as a wave departs and another returns, the grain of sand is changing its state even as we watch. As the chaos theorists have pointed out (using something like this very example), there is no accuracy if you're willing to look closely enough.
Thus it is with elections.
So, what conclusions do we draw from this? First, elections are necessarily indeterminate, especially on a national scale. Second, "the outcome" of an election is, like a map of a shoreline, determined by conventional rules for assaying the outcome. You don't have to map every grain of sand, and you don't have to count and re-count and re-count. But the rules also say that the outcome isn't always what the first count says. The rules--embodied in laws and practices--specify when a recount is required or reasonable. For example, Florida's laws say a recount is automatic when the percentage difference is less than 0.5%. And Florida's laws say that a county electoral commission can mandate a hand-count when there are signs that looking at each ballot might change the result.
Fine and dandy. Unfortunately--and this is the third conclusion--the rules themselves are subject to indeterminacy. The closer we look at them, the less clear-cut and accurate they become. And the rule for determining the rules is a court system that affects the very phenomenon it's adjudicating; becoming president on the basis of some cheesy court decisions alters the nature of your presidency.
Fourth conclusion: KM systems that equate knowledge with accuracy and intelligence with correctness are going to miss much of what's valuable to know and share.
David Weinberger is editor of Journal of the Hyperlinked Organization.