I think this is very important step. If this works, then we will have real data to argue about providing free basic internet connectivity (Enough for text for example) for entire world population.
This would allow for text only MOOC's accessible by everyone at the very least.
I think you are right to point out the value proposition to a text-only internet. But the implementation risks are probably more second-order and worth considering.
The contra case is that someone will twist this into a model to charge <up> for everything else. its better to create better public goods, than bargain for special access to private assets, when the quid-pro-quo will surely be some form of reciporacal monopoly rights (ie, special interest regulation).
This would allow for text only MOOC's accessible by everyone at the very least.