First, there are clear cases where one technology is better suited to a task than another. I'd no more write long papers on a cell phone than I would carry around a laptop to use as a personal communications device. However there's a large chunk of tasks where either tool will suffice, and which "should" be used is more a factor of the local conditions than the features of any one technology.
Secondly -- and more importantly -- this discussion is tool-centric. We have a hammer (two, in this case) and are going around the development landscape searching for nails we can drive home, and it's a race between the two hammers to see who can hit the most nails. This is inherently the wrong way to apply ICT in development.
We shouldn't be arguing about mobiles vs computers, or even OLPC XOs vs Intel ClassMates, or Windows vs Linux, we should be arguing about specific problems in development, what tools could help, how, and for what costs (training time, implementation and infrastucture gotchas, as well as equipment costs).
You can't really argue with this position, but of course I'll try: Sure, we need to pick tools that are appropriate to our objectives. BUT objectives themselves need to be appropriate to their contexts. And the critical contexts, the most critical contexts, share certain characteristics. I can probably run a pilot project that helps a few kids in a really poor country -- I'll go with Chad and "MDG education" for US $0.50 -- use laptops to write longish reports and build both skills and knowledge in the course of their writing. If I want to have impact across a wide swath of schools and students, however, I'll be better advised to focus on an objective that emerges out of Chad's educational context, perhaps increasing primary-school completion and lower-secondary enrolment, and that takes advantage of available infrastructure and of tools that are in use and supported.
You can walk this context/objective/tool-choice discussion back any number of ways. (For example, you could say that my objective, to have impact on education outcomes in a poor country, determines my characterization of context.) But IF we stay within the context of the "bottom billion" and the countries in which they live, I'll wager that we find out that laptops are rarely if ever the appropriate tool to achieve scaled impact in schools.