You don't really need math. But the thought processes learned through math training are really useful.

For instance, when a scientist publishes a paper, he could pay a fee to cover administrative costs. Then the article appears online, free to all.

That model might work fine in some instances, but not always.

Here in Mexico, it is actually quite difficult to get money to publish. Although the "Astrophysical Journal" has more prestige, readership and impact than "Monthly Notices of the Royal Astronomical Society," I know several people that submit papers to the second because it is free to publish.

On the other hand, the administrative overhead is smaller if the University pays for a subscription (that covers the journal administrative costs) than a system to transfer funds from the research grants to pay for each accepted article.