Checking the first characters

This is the check almost everybody actually performs, and it is the one that inspects the part deliberately chosen to be reproducible.

The cost of a prefix

Each additional character multiplies the work by roughly 32, since there are 32 possible symbols. A handful of characters is quick on ordinary hardware. Going further gets expensive fast, and matching an entire 56 character address is the thing the whole system assumes is impossible.

So the difficulty curve is steep, and the flat part at the beginning is exactly the part people inspect. Somebody building a lookalike spends a modest amount to match the readable opening and then lets the rest fall where it will.

Why the readable part exists at all

Because operators want a memorable address, and the only way to get one is to grind for it. That is why several markets have addresses beginning with their own name. The name is not registered and not protected. It is a purchased coincidence.

Which produces an unfortunate result. The characters that make an address feel recognisable are the characters that carry the least assurance, because they are the ones somebody paid to produce.

What people get wrong

They check both ends and feel thorough. Checking the end is better than checking the start, since the end was not ground for, but a lookalike matching both ends is still cheaper than matching the whole. And the middle, which is where the difference will be, is the part nobody reads.

They also treat prefix matching as a first pass that will be followed by a real check later. The later check does not happen. Once the opening looks right the page has already loaded.

The end of the string

The final characters are worth more than the opening ones, because nobody grinds for a suffix. If you are going to read part of an address rather than compare all of it, read the last twelve. That is a better bad habit than the one most people have.

What to do instead

1. Copy, never retype.
2. Compare the whole string mechanically, not by eye.
3. If you must read, read the middle.
4. Keep one trusted copy and copy from it every time.

None of this is clever. It is just refusing to substitute a cheap check for an expensive one, which is the habit behind most of the NEVER section.

A rough sense of the numbers

Treat these as orders of magnitude rather than measurements, and they are our own estimate rather than anybody's benchmark. Four or five characters is a matter of minutes on a laptop. Seven or eight is hours to days. Beyond that the cost climbs out of reach of casual effort and keeps climbing.

The point of the numbers is not the numbers. It is that the region people inspect sits comfortably inside what an ordinary person can produce over a weekend.

What would move this

A collapse in the difficulty of generating prefixes would make it worse, not better. Moving this statement out of NEVER would need the opposite, an address format where the opening characters were structurally constrained rather than freely grindable, such as a registered prefix issued by somebody.

That would mean a registry. Onion addressing exists partly to avoid one, so the trade is unlikely to be made.

• Every character weighs the same
• The 32 symbol alphabet
• The addresses