Further, enterprise software sucks for casual users, but is often superior for expert users than consumer software that is less featureful and robust. I think that depth vs learning curve is a bit of a false dichotomy, but complexity geometrically increases difficulty in making "intuitive" software.
I've worked in a few Fortune 100 organisations, and I've got to say that enterprise software doesn't tend to solve hard problems: financial reporting, workflow support, resource planning, HR aren't inherently hard.
What is hard, is selling them into big organisations, commitment of the management hierarchy, deployment into the processes and sometimes forcing end-users to deoptimalise their results. This all costs huge dollar figures, and the organisations need to sell that internally and come up with suitably vast business cases.
Maybe I'm being a tad cynical here, but I have participated in the process several times after being informed that a cheaper, more appropriate solution for these "enterprise needs" couldn't get buy-in - and in enough big 100k+ enterprises that it's not just unlucky happenstance.
Most of the software that solves big, hard problems are domain-specific, not really enterprise software. They are deployed for point- or black box solutions for the business: things like industrial process control, tax/logistics montecarlo tradeoff analysis, MPLS tag path provisioning, factory and warehouse logistics optimisation, geophysical seismic analysis for hydrocarbon prospecting, etc etc.
I wonder if Jason has ever worked with real enterprise software, either as a user or a developer. The reality that he seems to miss is that the enterprise software arena is full of big, hard problems that have huge dollar figures attached to them. But I suppose that's why Oracle is worth more than $100 billion; they're that "someone else".