Reuse of what, exactly? Selectors? Properties? My assertion is that if you need to re-use properties a lot then you should make use of a preprocessor framework and then use a combination of mixins or extends. Re-use of selectors is an unnecessary artefact of how OOCSS (and similar) suggest you should write your code.
> If you want to build a library of reusable components, IDs are obviously not what you want.
If you're building components that could end up on any page entirely outside of your control, I agree. But: horses for courses. It still has valid uses, and avoiding it entirely isn't necessary.
> Secondly, if you use IDs which were added for JavaScript or fragment links, you'll introduce some coupling which really shouldn't be there.
You're looking at this backwards: IDs exist for many purposes, not just CSS. Selecting elements by ID in JavaScript, or using a fragment identifier to link to within a page are just two other examples of how they can be used, and not considering the other purposes of the attribute is part of the problem (see also ‘CSS classes’).
> Same thing with JavaScript. If you need a class, you should add a prefixed one (e.g. "js-foo").
How does the developer benefit from writing the classname like this? Why is a prefixed classname any more informative to you than a selector that tells you a little about the context in which the element appears? I've actually come around to the view that if you're using this class to select on any other criterion than the semantic purpose of the element is wrong, and it would be more appropriate to use a data attribute to contain the hooks for additional behaviour.