Software estimating is not a particularly pleasant task - neither for developers (they generally have no training in the area and so don't even know how to approach the problem) nor managers (who have too little technical knowledge).
And since 95% of development is writing in-house CRUD stuff or putting together lightly trafficked web-pages, it's tempting to bask in ignorance and pretend that not only is the task almost impossible anyway (and so pointless) but that in-fact it's unnecessary. And for IS department mickey-mouse projects, this is probably the case.
Some of us however have had to pitch fixed price tenders for large software development projects, for example. And no, the customer isn't going to accept that it will be done when it's done and that we should just start a few sprints and see how it goes. And if you get the estimate wrong (too low), you screw yourself and lose money on the project - too high and you don't even get the contract. In this environment, you learn quickly that estimation is vital, difficult but possible with the appropriate effort. And yes I hated doing it the first time but it became almost satisfying after a few iterations having gained some confidence.
This assumption that a "post-industrial", agile, folksy-anecdotal approach is appropriate for all software development really pisses me off. It breeds developer myopia, ignorance, self-importance and exceptionalism. Believe it or not, engineers have estimated, planned and executed very complex systems inside and outside of software. But it's only in software engineering do you find clowns earnestly claiming that estimating or planning developing some crappy CMS system is beyond human capabilities because software is so "different" and that estimating and planning a project like the Hoover dam was easy in comparison.