How long each address is

Take one of the strings printed above, remove the dot and the suffix, and count what is left. You get 56. Do it to the next one and you get 56 again. Do it to the third and the answer does not change.

That is the entire computation behind the first tile, and it is worth doing once by hand because it makes the rest of this board legible. Nothing here is more sophisticated than that. The middle tile draws the same three counts as bars, which is why the bars are flat, and a flat bar chart is more honest than a chart scaled to make small differences look dramatic.

The length is not a style decision. An address in this format is a fixed amount of binary written out in a restricted alphabet, and the amount of binary is set by the design of the format rather than by whoever generated the key. That means every address of this generation, everywhere, from any service, is exactly this long.

The practical consequence is small but useful. Length is the cheapest sanity check available. If something you were given is shorter or longer than 56 characters, it is not an address of this kind at all, and you have caught a problem in about two seconds without needing to compare anything.

The consequence people over-read is the interesting one. Because the length is fixed, matching length is completely uninformative about whether an address is the one you wanted. Every wrong address you could possibly be handed is also 56 characters long. A check that everything passes is not a check.

The third tile adds the three lengths together and gets 168. That is the total amount of text this site holds as external data. Everything else on the board about the strings is derived from those 168 characters.

It is a small number for a whole website to be built on, and stating it plainly is part of the point. Most reference sites about a market are built on a much larger and much vaguer input: things people said, screenshots of unclear provenance, other sites. This one is built on three lines of text supplied once, and when it runs out of things to say about them it says nothing rather than reaching for a fourth source.

• It rules things out fast. A string that is the wrong length is wrong, full stop, and you do not need to compare it to anything.
• It sets the size of the checking problem. Whatever comparison you do, 56 positions have to be right, and there is no shortcut that reduces that number.
• It tells you how much screen you need. A string this long wraps on most phones, and a wrapped string is where transcription errors hide, because the eye loses its place at the break.
• It tells you nothing about correctness, reachability, ownership, or age. Those need something the length cannot supply.

The fourth tile is a different measurement entirely: it shows how much of the allowed symbol alphabet the three addresses between them actually touch. It sits on this panel because it is the first hint that the strings are not words. A word of this length would use a handful of letters heavily and ignore most of the alphabet. These do the opposite.

That property is explored properly on the panel about the alphabet. It appears here only to make the point that length is one axis and content is another, and that a check on the first says nothing about the second.

Does the dot onion suffix count?

Not in these figures. Every count on this board is taken after the suffix is removed, because the suffix is the same on every address in the world and counting it would add a constant to every number.

Could an address of a different length still work?

An older generation of addresses was much shorter. Those are not in use here, and none of the strings on this board is one.