How far apart the three are

Take two addresses of the same length, walk both at once, and count the positions where the characters are not the same. That count is the distance between them. It is the simplest useful comparison there is between two strings, and it maps directly onto the thing a person cares about: how many characters would have to change to turn one into the other.

The first tile shows that count for each of the three pairs. The second shows how many characters each pair shares at the front before diverging, which is the same measure taken from the opposite direction.

Every pair differs at most of their positions. The lowest of the three is still very high relative to the 56 characters available, which means no two of the published addresses resemble each other beyond the shared opening.

This is the expected result, and the panel says so plainly rather than presenting it as reassurance. Independently generated keys produce independent strings. Anything else would need explaining.

Imagine one of the tiles read very low: two addresses differing in only a handful of positions. That would be a genuine finding, and an uncomfortable one, because a near miss is exactly what a string designed to be confused with another looks like.

Near misses are dangerous in a specific way. A person checking an address compares fragments, and two strings that differ in three positions out of 56 will pass every fragment check that does not happen to land on one of the three. Distance is the measure that catches that, and it is the reason this panel exists even though its answer is boring.

It is worth carrying that idea away from the board. When you are handed a new string and told it is one of the set, the question is not whether it looks right. It is how many positions it differs from the one you already had, and a small answer is worse than a large one.

• It cannot say which of two addresses is correct. It only says how far apart they are.
• It cannot be computed against anything this site does not hold, and this site holds exactly three strings.
• It says nothing about what is served behind either address, which is not a property of the string at all.
• It does not scale to a judgement. There is no threshold above which a distance is safe, because safety is not what is being measured.

The second tile shows, for each pair, how many characters match from the front before the first disagreement. The answers are all the same and all small, because every pair shares the same chosen opening and then parts immediately.

Set side by side with the distance tile, the two make the shape of the set obvious: a short identical head, then near total difference. That shape is the reason this board keeps returning to the same practical point, which is that the useful part of a comparison is the part after the head.

Is a bigger distance better?

It is more ordinary. Distance is a description, and the reason to compute it is that a small distance would be a warning rather than a large one being a reassurance.

Can this compare an address I was given elsewhere?

Not on this site. The metric only runs over the three strings held here, and nothing is ever fetched or entered.