That is true, coding style can be differrent per use case.

As an actual user of applications I can tell you that I don't like slow applications and I don't even care about the code behind since it is secure and works.
We, as programmers, tend to see our code as something we give/share with other programmers and we forgot that someone else will use it at the end.
Someone who will lost time of his/her life waiting for something to finish... because we have a beautiful but slow code :)

I think it is more interesting when presenting facts to trigger some thoughts and point out some possible mistakes - we all do mistakes.
Posting just a solution is fun but is more fun and productive if someone read it and do some critics. That's the way to get better.
Making a jsperf comparison is a good way of that - nothing more!

Thank's for your respond and hope I didn't trigger any "weird" feelings to anyone with my posts - after all, in coding and war everthing is fair :D

PS. at the banking sector there was not a single time that I heard "that's slow but anyone can read it, let's keep it" - just the opposite! :)

Well, that's the question of the goal you are writing code for. Speed may be the critical parameter in some applications, and may be absolutely not important in some others. The code readability and, consequently, supportability by other engineers may be critical in some other situations. Learning and practicing complex language features produce different code altogether.

CodeWars has two ways to "support" a particular solution - "Best Practices" and "Clever", and you can choose which one is appropriate.

Posting the solution that is best from your point of view is a good way of promoting your point of view.

Posting claims that the existing solutions are bad in one way or another, is a good way to make yourself look weird. :)

I see a lot of solutions using "map" etc... bad although it is eye catching they are even 40%+ slower than my "old-school" "for" method.
Should the "best" solution be the "eye catching" one or the fastest one? (given both are working correctly)

(Some solutions are also altering the MORSE_CODE table - bad practice?)

Some tests in Chrome and Firefox:
http://jsperf.com/morse/2

With a very small optimization, the decode2_opt is the fastest one by far.

Agree, though it is still slower in Firefox and IE11.
I am sure it will be someday the fastest solution.

I just tested it in Chrome 40, and map is now slightly faster than using a for loop.