You either love it or you hate it, and the physicians who hate it typically hate it because of what you said - they're suddenly having to type their own notes into a computer rather than dictate it to someone else.
Personally I'm of the opinion that doctors doing their own data entry is better for liability and malpractice reasons, but I empathize with your wife and people who hate doing it.
Instead, the EPIC complaints are about the UI and workflow, which was modeled after the typical paper H&P (history & physical) forms clinicians would traditionally fill out. It comes down to the same UX lessons we've learned about most offline-to-online form transitions: directly porting a form is rarely the answer.
The anti-note sentiment has to do with the fact that they're copying-and-pasting from previous notes (often the vast majority of a note's content is pasted in), they're using MS Word to draft the notes because of a buggy editor UI, and then when they're in a situation they actually have to read the note, they're sifting through a pile of duplicated, pasted content.
There's a ton of opportunity for change here. There's an equal amount of risk involved, and selling these types of products into healthcare is about as fun as removing your fingernails. That's the piece EPIC does well, and why they're still a behemoth, even if their product is a five out of ten.
First, he'll spin away from you on his stool. Then touch a keyboard with his boogerhooks to re-log-in. (A shared KEYBOARD in a room of transient often-sick people ... did you ever do the biology experiment in high school where you swab surfaces and inoculate a petri dish? The keyboards came out worse than the toilet seats.) Then he'll mouse around on a bunch of screens showing inconsistent or at the least confusingly presented data (listing, for example, all medications you've ever been prescribed with no obvious way to highlight which ones are current). Finally he'll find the right text box for the notes.
As your doctor types in this text box, various auto-completes will come through. Your doctor will then choose from these auto-complete options, which generally make a mockery of the English language and basic grammatical agreement (gender pronouns, etc.).
The idea that Epic has empowered, or even in any benign way nudged, doctors to begin doing as good a job of data entry as they once did of dictation and charting is just complete hooey.
I have also personally had a physician using Epic Hyperspace enter an incorrect immunization for my then months-old child. This doctor in fact had the correct immunization schedule for children memorized and could have written it out longhand perfectly. But it wasn't until the next visit that they detected the out-of-order sequencing of the immunization series. This was perhaps not the fault of Epic, exactly, but was certainly mediated through Epic Hyperspace and it did absolutely nothing to prevent, flag, or otherwise mitigate this completely detectable, violation-of-deterministic-formula situation. (Everyone was fine and no harm ... this time.)
Way off the original topic, here, but let's just say regarding Epic -- whatever it is that's driving adoption of that system is not naively comprehensible to those of us coming from tech/software. It's not quality, cost, UI/UX, safety/accuracy, interoperability, standards, network effect, viral effect, or the rest.
Yet, whatever Epic is doing (or whatever esoteric motive their buyers have), it seems to be turning out well for Epic.
So attempting a tie-back to the OP -- it's not clear that long-termism of a business necessarily has any salutary effects on the rest of the world.
The idea is great, but it just runs into an absolute brick wall of realities of the health care system and human behavior.
The problems with EMRs are really just a symptom of a real problem in medicine which that every department head wants to do things their own way. And there isn't really a good way of selecting which is best amongst the various processes/methods. The more prestigious the institution the larger this problem typically is.
And I used "good" to damn with faint praise. It generally meets customer requirements, and probably doesn't cause as much iatrogenic harm as whatever systems it replaces. For what it costs, I would personally expect quite a lot more out of it than it delivers.
But there's only so much you can do with a perpetual parade of recent college graduates that generally turn over after about 2 years.
This would be news to my wife (a surgeon), or any one of her coworkers. The software is universally viewed as terrible by everyone I've talked to. It actively results in worse records because (at least as configured) it's so cumbersome to use that many notes are frequently kept out of the system. A management plan that takes seconds to explain verbally requires minutes of data entry.