said by mouse :

We are running a website that provides for a password forgotten service. A new password is being generated to get the customer back in. Unfortunately some of the passwords are difficult to read/interpret with certain letters/digits looking very similar (1,l,L,i,I,J,0 and O) especially depending on the font used by the recipient.

Just wondering how difficult it is to change the output for password generation to exclude those letters and digits?

said by yock :

This seems like the wrong solution to me. Why not simply recommend through your documentation that users use serif fonts?

said by Steve :

Which would you rather provide tech support for:

a) forgotten-password link that avoids troublesome characters
b) "How do I change the fonts in my UnknownMail client?"

Corollary to (b): wanna teach my grandmother to change fonts on her browser?

said by nixen :

Two possibilities: 1) set your display page's FONT attributes to instruct the user's browser to render a Serif-based font; and/or, 2) put instructions in to use cut-n-paste to load the temporary password into the client's connect info.

said by Steve :

The former is not universally reliable, and the latter is just more stuff to support.

said by Steve :

Friedl's rule: the best tech support is that which is not necessary

In 1960, DEC engineers realized that in specifying connectors on a frame, where numbers mark the card slot locations and letters mark the connectors on individual cards, some letters cause confusion. Thus the letters G, I, O, and Q were dropped to avoid confusion with C, 1, and 0. The remaining 22 letters were since known as the DEC alphabet[citation needed].

said by yock :

This seems like the wrong solution to me. Why not simply recommend through your documentation that users use serif fonts?