Study 1 of 8 in Mistakes
Mixing characters from two addresses
The most specifically plural mistake there is, and it produces silence.
The question
How does a reader end up with a string that is part of one address and part of another?
The three WTN mirror addresses
hn2paw7zfvndw3dovycegeqmvvnf4pl67b3g2p7pohjlzavloosh73id.onionhn2paw7zrgujyhnt6mgxlt2q6uhgbke4itpqitxhyfbumq3wtnckbuyd.onionhn2paw7zadwkcra3qzv5e4q547i7e5lvxm62cfxqftuqdu7moiu2ceyd.onion
Set out flat, in the arrangement they were handed over, with nothing added. Nothing on this page marks one of them as the main one, because on this site that would be the one claim the studies are all arguing against. This reference opens none of them and measures nothing about them.
How it happens
- Two entries in a note lose the line break between them and become one long line.
- A copy is made by eye from a screen, and the eye returns to the wrong line partway through.
- A stored string looks wrong, so a reader corrects a few characters using another entry as a reference.
- Two windows are open, and a paste comes from the one that was not intended.
Why the shared opening hides it
All three published strings begin identically, so the first several characters look right no matter which two you have crossed. A reader who checks the beginning, which is what people check, sees exactly what they expect.
The shared head study makes the general argument. This is its most concrete consequence: the part of the string that reassures you is the part that cannot distinguish the members of the set from each other.
What the result does
Nothing. A string that is not an address produces silence, which is identical to the silence produced by a route that is not answering, as what silence cannot tell you explains.
So the reader concludes that a route is quiet, tries another, and never discovers that their stored copy is damaged. The damage persists and will be met again.
The second order damage
The reader now believes one of the three is unreliable. They may prune it, which is the habit warned against in quiet and then back, and they have removed a working route on the basis of their own transcription error.
A set of three quietly becomes a set of two, and nobody made a decision.
Why partial correction makes it worse
The third cause in the list above deserves emphasis. Repairing a suspect string character by character, using another entry as a guide, is the most reliable way to produce a hybrid.
The correct response to a suspect entry is always to replace the whole line from a fresh copy, never to edit it.
How to make it hard to do
- Store the three as separate lines in plain text, in something that does not reflow long lines into each other.
- Copy with the buttons rather than by eye. Fifty six characters is well past the length at which people transcribe accurately.
- Never edit a stored entry. Replace it whole.
- When comparing, read from the end backwards, because the ends are where the strings actually differ.
How to detect it after the fact
Compare the stored entry against a source character by character, ideally from the tail. A hybrid will match at the beginning and diverge somewhere in the middle, which is the pattern to look for.
This is one of very few checks in this whole subject that gives a definite answer, and it is available precisely because you hold the set to compare against.
The summary
A crossed string looks right at the start, fails silently, and gets blamed on the market. The set makes it possible and the set is also what lets you catch it, if you compare rather than glance.
The reason this one deserves the most care
Most errors in this section produce a bad decision that a reader can later reconsider. This one produces a corrupted stored string, which is a permanent object that will fail again and again while being blamed on something else each time.
It is also the error most likely to make a reader prune a working route, which compounds it. A single transcription slip can therefore end with a set of two, a false belief about reliability, and no way of noticing either.
Copying with a button rather than by eye removes all of that for the price of a click.
If there were only one address
There would be nothing to cross. A single address can be mistyped, which is an ordinary error with an ordinary cause, but it cannot be assembled from two sources that both look correct at the beginning. This mistake needs a family of strings that share an opening, and it is the purest example of a plurality specific error.
What this study is not. It is not general advice about typing addresses, which is a problem a reader would have with any number of them.