Symbols the eye trades

Almost everything on this board is arithmetic with no opinion in it. This panel is the exception, and it says so at the top rather than hiding it. The metric counts symbols that belong to a set this site chose: the round ones, the tall thin ones, and the two digits that sit closest to letters in most typefaces.

Choosing that set is a judgement about human vision, not a measurement of the strings. A different set would produce a different count. The count itself is exact, the boundary of what gets counted is not, and that is worth stating plainly before anybody reads too much into the tiles.

The worst offenders are already gone. The alphabet excludes four digits, and the excluded four are precisely the ones that collide hardest with letters. Whoever designed the encoding clearly thought about people copying strings by hand.

What remains is a second tier of confusion: letters that resemble other letters rather than digits. Those cannot be removed without shrinking the alphabet and lengthening every address, so they stay, and the reader absorbs the cost.

The gauge puts the total across all three addresses as a share of the 168 characters in the set. It is a substantial fraction, and it should be, because the traded set includes several of the most common letters in the alphabet.

That is exactly why the number must not be read as a warning about these particular addresses. Any string of this length in this alphabet would score similarly. The number describes the alphabet and the length, not the operator, not the market, and not the quality of the strings.

What it does justify is a change in method. When a meaningful fraction of a 56 character string consists of symbols your eye is willing to substitute, comparing by eye is a weak procedure. Not useless, but weak, and weak in a way that gets worse the more tired you are.

The third tile cuts one address into seven blocks and counts traded symbols in each. The line is uneven, and the uneven parts are where a manual comparison is most likely to go wrong.

Do not over-plan around this. The clustering is different in each of the three addresses and would be different again in any other string. The useful takeaway is not the shape of this particular line but the fact that such clusters exist at all, which means the difficulty of reading an address is not spread evenly along it.

• Copy rather than read. The copy button removes the entire problem in one action.
• If you must compare by eye, use a fixed width font. Proportional fonts make several of the traded pairs nearly identical.
• Compare in short blocks and take the blocks in a fixed order, so that losing your place is obvious rather than invisible.
• Do the comparison when you are fresh. This is the one variable you fully control and it matters more than any property of the string.

Which symbols are in the set?

The round ones and the tall narrow ones that trade with each other, plus the two digits nearest to letters in shape. The exact membership is this site judgement and is not a standard of any kind.

Would a different font fix it?

A fixed width font with slashed and dotted forms helps a lot. It does not eliminate the problem, because the pairs remain similar even when drawn well.