Would you just grab the latest 0.1 release of the hot new database and throw it up on your production systems because everyone keeps talking about the promise of it? This is a real concern at the engineering levels.
One of the problems with distributed generation is not that it eats away at grid profits - there is still profit from maintaining the grid that allows you to have power even if your solar panels break. There are price structures that allow this to be profitable for utilities (however everyone involved needs to look at it with fresh eyes -- homeowners need to stop thinking about it in just profit from unused kwh and utilities need to stop looking at it in terms of kwh pushed). The problem is that huge and expensive amounts of infrastructure are designed an manufactured with a one-way power flow in mind. Protection schemes are predicated on a star(ish) topology of distribution lines, where the assumption that power flows from the "hub" out.[1] Things like transformers are highly tuned for this type of flow, to the point where even small changes or running out of spec can seriously degrade the expected lifetime (usually thought out and planned on the time-frame of 30 years.) Power flowing the other way is a "break the line" event, period, to protect that equipment. A lot of substations don't have the type of capacity or equipment to handle power swings too far out of spec, because they designed that way for maximum efficience of the push model.
Something this article misses in it's analysis, is that power companies building infrastructure is absurdly expensive - getting right aways for power lines is a decade long, legally perilous, and highly prone to regulatory whim, endeavor (the NIMBY crowd is strong here). As such, once one is put in, the cost is amortized over 30+ year timelines, and efficient operation is actually strongly considered. Demand response and efficient appliances are actually in the interest of utilities, and they do recognize this. If they can't build new infrastructure to increase supply, they want demand to remain within their capability to deliver (and when done right, the base "connected" fee to the customers can be as or more profitable than more power to fewer customers).
Finally, all the understanding and models of how electricity works in an interconnected grid is based on models assuming "roughly" steady state capable generators. Things like solar and wind have some problems in this, as they can contribute instability to the overall system - if the wind dies, you have to have hot standby power to keep voltage levels up. If everyone in a region has PV, and clouds move in, they will be demanding more power from the grid. On a "partly sunny" day, this result in weird spikes, again placing strange wear patterns on transformers and generators that are ramping up and down in response.
Even the engineers I know who are all about this stuff are hesitant to just deploying it, because of all this technical challenge. Just like I would be hesitant to throw an unknown datastore into a stable working system without staging, testing, and otherwise slowly integrating it. It's even worse for the utilities, because they are in a "damned if you do, damned if you don't" position. What is worse - rolling slowly with the new technology and being blasted as obstructionists, or going head first, and being blasted when the unexpected happens and breaks a bunch of things, or finding a middle ground and making everyone unhappy?
[1] at transmission levels, this isn't true, and in dense urban centers the topology is meshier, but this statement is still true for a (geographically anyway) majority of power distribution.