Logic or Execution?
I've seen the best mappers make and praise the practice of making maps that are expected to be solved using difficult flings and box-handling techniques.
This could range from just using momentum in a previously untaught way (e.g. bouncing a laser cube with portals to hit two laser catchers) to being borderline glitch exploitations (e.g. standing half in a portal on a low ceiling and shooting) Maps that rely on parkour like this and ones that require quick reflexes especially garner unconditional love from some people and burning fiery hate from others.
On the flipside of this coin we have logic challenges, where more emphasis is put on thought than action, and it's the dynamics of test elements that solve a puzzle. These range from the ultra-straightforward (VALVe's entire SP campaign) to the incredibly complex (see my Laser Circuits for a good example). These ones too can frustrate some and astound others, but here because all the tools for the solution are made clear, it's fitting those tools together that makes a puzzle.
I've noticed that there is a distinct divide between people who like either of these solution methods. I am in the latter category, but that's just my opinion, apparently. I created this thread and poll to truly gauge which is the more popular problem-solution to this problem-solution problem.
Did you know that American English isn't even my first language? It's actually [spoiler]British English[/spoiler]
Released maps of note:
[spoiler]Versatile Diversity
L-Circuts[/spoiler]
I prefer logic, but I like Josepezdj's maps which are both logic and execution, because his maps and greykarel's maps are at least not about trying again and again on the same thing like a room full of timed buttons that need all to be pressed within doing weird tricks...
@Antivector: be fair and add a 3rd option: BOTH 
useful tools and stuff here on TWP 
[spoiler]
[/spoiler]
I far prefer logic to execution. Sometimes I put a *bit* of execution difficulty, but in my opinion a good map is at least 90% thinking and no more than 10% doing.
If execution maps have a special trick, (not straight foward and too hard) together with logic, it's a good map too. Also, what srs bsnss said is true.
I personally prefer logic solutions.
My reasoning is much like VALVe's. It is incredibly painful to KNOW what to do and just not be able to pull it off. That is the type of thing in a map that instantly turns me off. I'm not saying some things shouldn't have some hard execution, maybe changing a portal around a few times for the perfect fling, or something similar, but my rule of thumb is if you can't do it yourself 4/5 times, as the actual maker, then there is a problem 
.
This. I tend to like to make puzzles easy, but fun to solve. If it isn't fun, then it isn't a good puzzle. That said, I'm not a good designer of "hard" tests, but I try. I like to show the player a basic type of puzzle earlier on, and then later in the map expand on that same mechanic.
That would've defeated the purpose of the poll, though. I wanted to know which method the Portal 2 community wanted to emphasize on. I know the methods aren't mutually exclusive, but people do prefer one style over the other, and I suppose the consensus is a preference to logic. That's what I wanted to find out.
Did you know that American English isn't even my first language? It's actually [spoiler]British English[/spoiler]
Released maps of note:
[spoiler]Versatile Diversity
L-Circuts[/spoiler]
I definitely prefer logic, but sometimes it's OK to add a little bit of time-based difficulty to a puzzle to make it more interesting, or to do something that would not be otherwise possible.
The things I try to aim for regarding timed operations in my own maps are:
- the player can work out exactly what they need to do before starting the timed part of the test,
- the time limit is not unreasonably short and
- the cost of repeated failure is low. (IE no death and no laborious reset procedure.)
For example:
example / spoiler
Other Portal 2 maps: Medusa Glare
Portal 1 maps: Try Anything Twice | Manic Mechanic
I think I'm with HMW here, I definitely prefer maps that I can think out, but it's fun finally getting the solution, and then doing a cool maneuver to execute it. I think a balance of both (or should I say a combination? ) makes the best map.