You don't really micromanage things, just like managers in a company shouldn't micromanage things. You don't make decisions on how much bread should be made by each of the 300 bakeries; you decide that making bread itself is important so you give subsidies. You decide that making bread locally is important so subsidies are only available to locally made and not imported bread. You decide that whole wheat bread is more important than white bread so you give subsidies only to whole wheat bread.
First, as you say, because of communication bottlenecks. 300 small bakeries (number made up) keep a better eye on how much bread is needed in Manhattan than a technocrat in DC, or even Albany. As you say, we now have the communication technology to overcome that.
But second, there is the command problem. You may be able to get the data there, but who's going to make the decisions? They were made by 300 (or however many) bakery owners; now they're going to be made by one or a few people. They may have the data, but do they know what to do with it? Do they know enough about bakeries and bread? Do they have the mental capacity to replace 300 people?
This gets worse as you get bigger scale. What if you're not just trying to do the bakeries of Manhattan? What if you're trying to do all the food supply in Manhattan? All of retail in Manhattan? All of retail in the whole country?
The other problem with a commander is that they can decide that they want something, whether or not the data supports it. And their subordinates may decide that they'll get better promotions (or at least keep their jobs) if they 1) do what the commander says and 2) tell the commander what he wants to hear.
Better ability to communicate the data does not fix the second problem at all.