In general, mathematics papers have no first author; instead, authors are listed alphabetically. Therefore, citing “Smith et al.” is at best misleading (it implies that Smith did all the work) and at worst insulting (Zelmanov never does any work!). Basically, all the authors are considered equal but citing “Smith et al.” messes this balance up.
The “et al.” style is made slightly more complex because it means that the “big shot” may come further down the list. For example, Zelmanov is the big name but the paper is Smith, Smythe and Zelmanov, cited as Smith et al. So you don’t realise that it’s a Zelmanov paper.
My question is:
What should be done if a journal changes your citation of Smith, Smythe and Zelmanov  to Smith et al. , especially when Zelmanov is the big name?
Is the answer simply to ask them to change it back to Smith, Smythe and Zelmanov? I think that changing the citation to “Zelmanov et al.” undermines the equality ethos even more — it is a slap in the face to Smith and Smythe, and anyway how do I not know that Smith won’t win a Fields medal himself one day?
I should say that before today I have never seen this in a maths paper.
Added 2018: This was resolved nicely. I send the copyediting team an email asking if every instance of "et al." could be changed back "especially on line 244 because "et al." here is just a single person!", and if not can I edit the certain bits to remove the names. So they removed each et al. and it appeared online a week later. No hassle (but lots of worry!).