Anyone who has worked on any sort of sufficiently large and complex application knows that this isn't true. Programmers come and go, people have different styles, requirements change, deadlines are imposed, etc. There will always be grunt work, no matter how good a programmer is.
My core strength is systems administration and building resilient infrastructure. That's my 100% skill.
The development stuff is just what I'm doing for my side projects.
I dev to 95% minimum-viable-product state and then launch and look at the reaction and shake out the bugs. If the service has no traction, I move on. If it has a customer base, then I slog through the tedious 5% (that takes just as much time) to try to perfect the product... but if I don't need to do it, I won't, because I find it too mind-numbing.
I have a co-worker who would rather write tests for code on a project he doesn't work on, then write the feature the test is for. He just loves it.
I'm not the parent, but I'll answer, from my own experience.
The "final tasks" of a project aren't important, in the grand scheme of things, at least, not the way that I work. If I have an idea, let's say "Github" for example, what's the core of the application? Whatever your answer, I'd bet you dollars to doughnuts that it isn't "the login page", or "the frequently asked questions", or "what color the navigation bar is".
The 'core' of the application is the interface between git and github. That's also, generally speaking, the 'hard part'... and the developer that can take you from a blank screen to a demo, by his or herself, and can deliver that product to market, is the ideal developer. If you can hire a guy that can take you 95% of the way there, and then outsource the fiddly bits like login and navigation to oDesk or whomever, then you've got a real talent. That is the type of developer the parent seems to be describing, at least to me.
As to his self-understanding, I dramatically prefer working with developers that know where their limitations are, and know where they're not good. I find myself getting bored with the result of a QA report, and it doesn't excite me. Solving a problem excites me. Tweaking an animation to prevent flickery scrolling on IE7 does not excite me. It took me a lot of years to get from being the 95% guy to the 100% guy, but even my 100%ness is qualified -- it has to be within a given set of parameters. For somebody to acknowledge their weaknesses, in my opinion, demonstrates a higher degree of self-awareness than the average person who doesn't.
> a guy that can take you 95% of the way there
The devil is in the details. You just quantified one of the most difficult things there is to quantify. What were your units for that percentage ? Dollars, hours, lines of code, functions, cpu cycles ? Language breaks down as soon as you do that.
I agree with you in spirit that if there's menial tasks that can be easily outsourced, then fine, outsource them. But as you used a model github to illustrate your point, let me use a model - airplane - to illustrate mine:
I have this idea, an airplane. I figured out that the Bernoulli effect can be created by designing a wing that has less area on one side than the other, thereby creating a vacuum and lifting the plane up. There - the "core" of my application (of physics, not of software) is done. I have a functioning prototype and a solid theoretical foundation for proving it works. I'm at 95%.
What's left is to outsource all this menial shit. My QA staff advised me that when two propellers vibrate at the same RPM, harmonics are created that destroy the airframe. They also found out the design fails in icy weather. Oh well, a subcontractor can install a heater. The current airframe design is so weak that it creates a condition such that the plane can take off but cannot immediately land, as the weight of the fuel will damage the wings.
Oh well, I've got it 95% there...
If your goal is to prove a market quickly, the 95% guy is definitely good enough. Sure, you'd take a 100% guy if you could get 'em, but they're even harder to come by. Anecdotally, I know quite a few of the 95% guys myself, and they're doing really cool things.
But what they did in their initial stake is establish that flight is possible. They took strips of wood and cloth, and turned it into a machine, that flew, through the air.
Yeah, if your plane crashes because of propellor harmonics, that's a problem -- but that's a problem beyond the initial implementation. If you've invented the first airplane, that's a big problem to have, but if you've already demoed a working aircraft to the world, you've likely already moved past the 95% into a proven market, if not having fully established the market, and you can hire a team of engineers to fix the implementation details.
Either way, you illustrate a very important point, which is that where the boundary is drawn is crucial, and that definitely gets my upvote.
Either way, if it's taking just as long to write the generic CRUD of your application as it took to write the application itself, either your application is quite trivial, or you're doing something very wrong.
In practice, for most people, the "last 5%" doesn't get added in until after deployment, if at all. Regardless, if you're proving a market, and the highest priority is in determining whether or not there's money to be made on a product at all, the 95% guy is an ace in the hole.
The GP even admits that yes, the 100% guys are better, but they are indeed far rarer to come across, and usually involves other trade-offs.
You might have an inflated idea of how good you are at this shit. If you are really good, you won't have much grindy boring work to do.
If you're not taking a major part in the final tasks of a project, how can you be so certain you are good at preparing it for the future?