The font store gives the option to attach a PDF specimen along with the promo images. Since this PDF will be publicly available, what to do with the font in it?
If I keep it live/editable the font can be easily extracted. If I outline it, then the text preview looks bad because the "Enhance thin lines" option in Adobe Acrobat is on by default. This option makes some characters look bolder than the rest, a kind of bad hinting.
Should I export a version of the font with a read-only embedding option for this purpose? Does this prevent font extraction?
And second, which embedding option is the most common? I used to put "Editing the document is allowed" and "Allow subsetting".
Don’t spend a single minute on any countermeasures that you could spend on doing more worthwhile things for your foundry.
Nobody goes through the trouble of ripping fonts from a PDF when they could just download them after searching on certain websites.
As a user, I find it really irritating. As a type designer, I won’t do it unless required to (for some reason).
There's really no reason to embed the entire character set of a font in the document when those characters aren't used, so yes, use subsetting. In addition, subsetting will prevent people from extracting the entire font from the file, which you're concerned about.
Converting the pages of a PDF to images (dpi choosable) is always possible. For machine learning it's the easier way to learn the shape of glyphs.
Subsetting is a usual method to reduce size. E.g. for the installer of Debian Linux supporting >70 languages only the characters used in the translation files are left in the fonts.
IMHO there is no need to have the full font in the PDF. Disadvantage is, that some users will see the full character repertoire or some special glyphs. I always look at long-s, \f, punctuation, numerical digits. And I want to see them in special combinations. That needs an interactive resource (rendering them as image in a webpage is ok.)
At the very least you need to show the basic alphabet, figures, and punctuation. For work in English, you don't need more than this for many uses (headlines, logos, etc.). But you don't really want to leave these basic characters out for a type specimen, even if you're subsetting. So with subsetting you end up leaving out less-used characters, which, for the pirate, may not matter.
The other thing is that things like kerning and OpenType features are left out when embedding a font in a PDF. When you extract a font, you get only the character outlines and the advance widths, and maybe the vertical metrics, not the full font. (At least this is my understanding.)
So if someone downloads the webfonts from the site, they don't get the full character set or in all styles.