Alexander the Great did the same thing in order to solve the Gordian Knot[1], he changed the spec by removing the complexity (in one stroke).
I was not in a position to even think about proposing the removal of any spot that had to be guarded at all times. Doing so, might have resulted in someone (including myself) doing double-shifts for a few days. Only he who has the power can do it (e.g. Larry Wall for Perl, Alexander the Great for the Gordian Knot and the sergeant for the military camp).
Not doing something, or doing something dramatically simpler, ends up with less code being written. The cost of writing the code is usually a small fraction of the cost of supporting it during over the life span of the system.
One problem I faced was a client that as part of an e-commerce re-implementation brought a huge list of requirements for promotions and how they should work. There was hundreds of use-cases, some being very complicated.
I came back to the client and said, "are you sure you need all of this?" After working with them to investigate, it turned out that they had just documented all the features of their existing system. In the previous ten years, they had created exactly one coupon, and that was for a flat 10% off.
That was the functionality we initially went live with.
Coding monkeys consider the spec the problem and try to write code that solves it.
Programmers consider the spec a part of the solution and evolve it as necessary.
So many times I've seen someone struggle to implement a poorly written spec rather than just say 'hey can we just do X instead?' If X solves the REAL problem then sure you can.
100% true! So many times I was assigned to the task in issue tracker, where requirements was written, estimation was made, but after I sat and tried to solve given problem, it turned out that 80% of issue description was false - real problem was totally different than it was written in task.
What seemed right on planning meeting, turned out to be quite false in the process.
But quite often managers/product owners etc. don't understand this. They often believe 100% in planning ahead and issue tracker/backlog descriptions... And when something is going off the track (like it sometimes does - when problem needs extra time for thinking through) they aren't so glad because thinking and design process was not budgeted in issue tracker (even if it could lead to the better design on the long run).
In the end, you provide value or you don't. The hardest thing about any product development is to stop taking things out or worse yet, keep adding things in. The top performers are usually the ones who can weigh these pros and cons almost immediately and force the product in a better direction.
> One epochal example not mentioned in Gabriel's paper is from distributed hypertext systems. Early distributed-hypertext projects such as NLS and Xanadu were severely constrained by the MIT-philosophy assumption that dangling links were an unacceptable breakdown in the user interface; this constrained the systems to either browsing only a controlled, closed set of documents (such as on a single CD-ROM) or implementing various increasingly elaborate replication, caching, and indexing methods in an attempt to prevent documents from randomly disappearing. Tim Berners-Lee cut through this Gordian knot by punting the problem in classic New Jersey style. The simplicity of implementation he bought by allowing “404: Not Found” as a response was what made the World Wide Web lightweight enough to propagate and succeed.
I remember a very specific example of this from Perl 6 development circa 2009. perlgeek (I think) and I had been struggling with coding up the sequence operator for a couple of months. Every time we thought we had it, someone would come up with a case that broke the system again.
pmichaud (not sure if it's the same pmichaud here on HN) comes up to us and says something like, "Guys, you're not getting this, let me do it." And I was kind of offended, because it wasn't like perlgeek and I were chopped liver; why did he think he could easily solve a problem that had been plaguing us?
But here's what pmichaud did. He went up to TimToady (Larry Wall) and said, "There's too much magic in the sequence operator. We need to simplify it." I don't know which combination of the two of them came up with the details, but less than 24 hours later the spec had been changed so that there was no magic on the right hand side of the operator and they had a working implementation. Any sequence possible before was still possible, but the user was required to be more specific about what they wanted. After the change the median sequence took maybe two more characters to specify ... but suddenly 90% of the hard part of coding the operator was gone.