Puzzle Producing Process.
Step 1:
Brainstorm!
- Write anything portal puzzle related on a piece of paper
???- If you want the chamber to start around a certain theme, brainstorm around a certain element. E.g. Lasers. -> lasers can turn on relays etc. But make sure that it incorporates another element with it!
- Talk with another person about your puzzle! Ideas can massively appear when you talk with someone else.
- (optional) Look at your previous brainstorm for missed ideas which you can incorporate in
Step 2:
Mix and match!
- grab your chosen puzzle ideas and try to match them together
- this can be using words or perhaps drawing diagrams etc.
- Too many ideas or elements in one chamber can be confusing or too much for a player. If you have a lot of ideas make sure to split up your chamber into smaller chunks (many chambers)
Step 3:
Draw your concept chamber!
- Do this by a birds-eye view or by a oblique perspective drawing on a piece of paper
- Draw your puzzle elements the same every time (like a key) so you can understand them when you get back to it
Step 4:
Make your concept!
- Start making your chamber in hammer
- Do not detail
- Only put walls, floor and puzzle elements
Step 5:
Play and refine!
- Once you play-test your own map, move/add your puzzle elements etc. until your puzzle is complete
- After you are happy with all the elements, this should be the part where you get someone else to play-test your map.
- If your play-tester finds something majorly wrong with the chamber, fix it!
Step 6:
Detail!
- This is where you can move things around in order to make the chamber visually more pleasant
- Add props etc. and make sure it doesn't change your puzzle
- This should usually take the longest amount of time
Step 7:
Release in beta mode!
- There are pleanty of ways to beta test:
???- This can be done by an early release and keep on refining the map in WIP from step 5.
???- or by getting another person(s) to (critically) analyse your map while they play it.
??????- Ideally getting around 3 people to test your map will crinkle out most-all the bugs / troubles they have.
???- Some other way
Step 8:
Release the final version of your map!
- Do it!
This looks so damn useful. Thanks mate 
My method is quite the same as LP's, I sketch the puzzles during boring lessons.
This is the core of non-linear puzzle making. Which is what people like. Well, the portal-thinking people, not the whee-a-rollercoaster-map-thinking people.
I guess a top-down design approach on pen and paper is the classic way to pretty much any level design task (I shudder to play a CS:S map with the mapper having no solid preconception of what he's doing). Here's the top-down view of the red room from my latest map.
On puzzling puzzles:
Lately I've began to have some serious thoughts on what makes puzzles in Portal puzzling. Here are my thoughts.
- Original solutions, if pertaining to some kind of real life logic, can be excellent ways to get the player to think (if not, but still cool, you will need to tutor players first). Remember the first time you played Portal? If I go back nowadays, I feel embarassed at how some of the puzzles took me time to solve. But everything was new back then.
- Dilemmas are your friend. A weighted button would cause a panel to rise, allowing you to fling somewhere. You have a cube for this.
However, you being this dick puzzle designer, it also raises a panel that would stop you in mid air.
- Allow for multiple ways to use your same elements, but only one (or a few) is the solution. This is rather tricky to design, but it is the key to proper red herrings. You don't want useless elements that are simply there to confuse the player. Portable walls that you don't use are okay red herrings. However, preferably a continous wall that is not useful rather than a singled out portable wall between black walls that looks like it HAS to be used.
- During the process of a puzzle, if you do certain steps and a player later has to break preconceptions of these steps, this will no doubt stump the player for a bit. What I mean is things like this: you need to get a cube across a fizzler, but at some point in the puzzle, in order to do this, you need to fizzle the cube (there's naturally a dropper in this scenario). Or, you need to establish a link of portals between points, but it's required to go through a fizzler at some point. Or, and here's one that was used in Patent Pending (and Mevious' map Traverse): a redirection cube is used to redirect a lazer into a receiver. Simple stuff. However, eventually the player realizes he/she needed to do that through portals instead.
- Continueing on the above example, think of a solution to your map. A logical flow of steps. They might be a bit obvious. They're fairly intuitive and the design hints at these through various means. You can rest assured most players will figure out what to do there. For example, an angled panel almost always means 'fling'. Then, carefully twist some of these steps so that it is still the end goal, but the 'how' is not as obvious anymore.
Here's an example I just made up with some of the above advice. Let's continue with the example of the dilemma with the button and the two panels.
Or, how about a light bridge instead of a panel barrier? Light bridges vertasile and lovely. You could portal this light bridge to gain height, to set yourself up for the fling. The solution would be to fling, and after exiting a portal, the cube needs to get off somehow. You need to push it off with the light bridge! Now, while flinging, this is sort of impractical, so then its a puzzle for yourself to get this to be viable. One solution is to simply make the room bigger, giving you more time while flying. Another is to make the place to portal the light bridge close to the panel, so it's easier to pull off, and to make the surrounding areas non portable, so it's hard to misplace a portal on something else. Pushing a cube with a light bridge is a bit hard, so an angled light bridge might work better. So, have an angled panel pointing down on the button.
Now, this solution is a bit obvious. Let's introduce another element: blue goo. An even more obvious solution would be: put the cube on the button, do the fling, then portal blue goo, and on that panel pointing down, to bounce away the cube. However, when the button activates, the goo stops! Now you need to use the light bridge again. The goo can still be useful in some other way: perhaps you need it on the bridge to gain enough momentum for the fling, or to bounce off of it when you land to elongate that fling.
However, the above example still feels a bit on the easy side to me, but remember that they will always seem easier to you, than to the player. The only way to properly verify this is to have someone test it.
Also, I have no idea whether the above example really works, but you get the drift, and that's where you test things first.
Again I pick 2 elements I want to make a puzzle out of, sometimes I add a 3rd. but then I try to make a challenge that uses each item atleast twice. and bam you have a puzzle.... the reuse of items in my opinion is key to a good puzzle
That article is a joke. It has to be. Most of the stuff in it is ridiculous. You can't treat a puzzle like a math equation. You will find that the best puzzle is the one you design for yourself. If you enjoy it, others will as well. Most game designers will tell this about game design itself, but it seems to work for level design as well.