Nothing gets programmers into more hot water more than the ‘keep it simple, stupid’ (KISS) principle.
There are 2 important ways it causes grief.
The first is when programmers assume it applies directly to them. They want to keep their work simple, so they want to keep their code simple.
The failure here is that there is always some ‘intrinsic’ complexity that is totally unavoidable as a byproduct of our physical universe and its history. If a programmer ignores some of that complexity and makes their code super simple, the complexity hasn’t gone away, it has just moved elsewhere. More often than not, it has gone directly to the users.
So, now the code is simple, but using it has become significantly more complicated than necessary. A super simple feature either doesn’t do much or it is horrifically hard for users to get real things accomplished. Either way, its value is limited.
If you want to keep users interested in using a system, then the code in it has to really solve their problems and it has to keep it simple for them. KISS applies to the solution, not to the way it was built. That is not simple code to write since it means ensuring that the users don’t need to lean on any outside resources. The system needs to know anything and everything about the problem, to keep it all up-to-date, and to never forget stuff, which is, of course, large and very complicated.
KISS can be applied to a specific sub-component of the mechanics, but it only works if you are avoiding some explicit overcomplication. That is, if you are thinking about adding stuff that will definitely never get used, then you can remove that. However, if it might get used someday, then by removing it, you are actually setting up a new problem. Maybe a small one, but it also could be a huge one in a short time from now.
The other way programmers get into trouble with KISS is that they assume that if they have to build a series of components, that making each one of them simple will help simplify the whole thing. The opposite is often true.
Simplifications, almost by definition are things that are not there. Gaps in what could have been. Those things might have been necessary or they could have been completely extra, but they are still absent. Collecting together a bunch of non-related things all together in the same place is another way to describe disorganization, and oddly, collecting together a bunch of missing things, related or not, is almost the same.
The growing absence of stuff is a form of complexity and if not handled explicitly it will get more complex. So, a small piece missing from one area can be worked around fairly easily, but when there are dozens of things missing all over, the workarounds are harder and more complex. Just remembering what isn’t there becomes a huge burden. So, if there are lots of things that are missing, their combined effect will be multiplicative, not additive, if they are not 100% independent (which they usually aren’t). Simple is not a cumulative property, complexity is.
“Everything should be made as simple as possible, but no simpler.” -- Albert Einstein
The key part of this quote is that while you can overcomplicate things, you can also do the opposite and oversimplify them, which is bad too. When KISS is misapplied or taken to an extreme not only does it craft a system that users hate, but it will also slowly degrade into a horrific ball of complexity that at some point becomes unfixable.
It’s always a shock to the programmers, particularly if they have been diligent about trying to simplify everything when it suddenly kicks back up and becomes the problem itself.
Stuff that is missing or not included is always going to be a big problem if you find out later that you need it. It’s that old ‘a stitch in time save nine’ saying. Filling the code with a lot of extra stuff that isn’t used ain’t great, but then it is a little better than waking up one day and realizing that what you didn’t do earlier -- when you had the chance -- is now going to derail everything.
Don’t overcomplicate things, but don’t use that as an excuse to swing way out to the opposite extreme, either.
No comments:
Post a Comment
Thanks for the Feedback!