I too have realised that I am guilty of Yak Shaving, having set out to create a video game I decided that none of the middleware that was available was sufficiently streamlined to boost my productivity by the order of magnitude that I required to be able to complete the game’s development in a matter of a few years. Without tools that were designed to harmonise with the way I chose to conceive my “worlds” I would take decades to realise my vision. Yet, the development of these tools would be made much more rapid through the use of a bespoke programming language – after all, none existed at that time that precisely fit my stringent criteria. I didn’t know myself quite what the specification of my language should be, at first… soon realising that I needed to engage in some Research in order to better perfect its Design. I read about the merits of separate compilation, Eiffel’s melting-ice technology and later on about live-programming. I read about functional programming, referential transparency, term-graph rewriting systems, symbolic programming and constraints. I thought a lot about how an incremental integrated development environment could allow me to build the language/API/system/tools around myself and given how similar this was to Smalltalk that a reevaluation of the GUI was not “off-topic” if its perfection accelerated my future productivity. Every detour could be justified this way and without any imposed deadlines on my hobby project I continued in this pattern, not even blinking as I spent two whole years redesigning my keyboard layout. I was having fun.

However, there was a darker side… having begun in 1991 it didn’t escape my attention a decade ago that my estimates were, how shall we say, a little “off”. I had become so absorbed in the immediate problem and so fascinated by learning about and integrating new technologies into my design that the original objective of writing a game within a few years having spent a majority of those years building productivity enhancing tools so that the game could be completed painlessly in half of the remaining time which would be a tenth of the time it would otherwise have taken had obviously failed. Numbers did not add up. I had taken too long and wasn’t even past the design phase. Yet, I couldn’t stop what I had been doing… I had to see it through without compromise. I had to strive for perfection as falling short of that lofty goal would determine the limits of my ability. How good could I be? I had to know.

And then there was the fear. The niggling doubt that festered at the back of my mind that cutting any corner now, avoiding any detours in a blind sprint to the finish, would inevitably welcome disaster as I would likely find my future self using my completed middleware unable to do some overlooked thing because I had “rushed it into production”. The longer I spent on it, the stronger this dread became.

Don’t shave your Yak.

At this point, twenty years in, I find it easier to keep shaving. Although, as I have shaved one end of my personal Yak (so to speak) and moved on to the other end, it has started to regrow its coat. Since I began there are new languages, paradigms and architectures to research, digest, assess, reject and/or incorporate. Is the task finite? Well, I am confident that I am finally ahead of the curve… new things remind me of old things, aspects connect and reduce, dissenting opinions are resolved with greater wisdom… does anyone need to stuff a mattress?

- Kyle at CivicPlus

What a great lesson (I keep saying I need to live more of my life like an agile project). Excellent job explaining how agile’s time boxing helps us focus to actually ACHIEVE things.

And now I have a wonderfully new, obscure expression with which to confuse my friends. Thanks!

So if you find yourself shaving a yak, you are not only many many levels of iteration away from your original goal, you are in fact not even doing the right thing, to get the thing you need for the other thing, for….