The testing mentioned in these articles are really a red herring in some cases. It isn't as much about the hamburger icon (or any icon), as it is about hiding data.

Having a button called "menu" performs poorly when compared to displaying the menu options at the top level. Likewise, the same "menu" button performs poorly vs an icon with young children who are still learning to read.

The common truth: If data is hidden, less people will bother to find it. Children are trained to poke around based on their intuition that something will be there (which is reinforced from past, consequence-free curiosity). Older folks hesitate to explore and tend to still function as if it were the PC era, where clicking on the wrong thing will break the system.

So the lesson of the day: Don't hide important things. Hide things that you're OK with your users missing.

Thirty years from now when the hamburger is commonplace and your grandkids ask what you were doing the revolution- what are you going to tell them?

I generally go for something like this: http://cl.ly/image/2Q2F2J3F2O3b

Ahhh the old argumentum ad consequentiam that strikes again the internets.

Use the hamburger, test it, and if doesn't, just analyze if it is the menu icon, the chosen navigation system or the product itself that is overall failing.

Don't skip some choices just because some article on the internet said 3 o 4 websites between trillions use it or not, or worse, they said it didn't worked well. There are many factors to be analyzed to conclude the sucess, like users are accustomed to using a system and suddenly another are. Years ago, I've heard a pseudo UX-guru saying that lazy loading lists never worked and the proof was an A/B test made on Etsy. That things hurt you worse than trying and take your own conclusions.

From the looks of it, the hamburger is going to be a pattern that everyone will understand because designers are willing it to be so. Personally I'm okay with it.

In user testing over the last couple of years, I've seen a definite increase in the number of people that recognize the icon and go straight for it on mobile. Desktop is another story.

Did fb discontinue the hamburger in their ios app? I'm seeing this in the current ios app:

http://imgur.com/IW2Y17U

Remember, there were times nobody knows the switch i/o icon. I think it's a similiar situation. But it takes time to get into main street.

One thing that doesn't seem to be taken into consideration is that every website differs. Like Aaron S. says, the tests are a red herring. They're tested on small websites in a quick conversion context.

To illustrate there are sites like Path and Facebook with their stream, immersive websites for say fashion brands, and information websites like wikipedia or government websites.

His advice (restructure IA for mobile, don't employ new interactions but use tried and true ones, let device manufacturers innovate and standardize interactions) are generalised and presented as best practices and solutions for mobile UX.

I disagree with these three particular points as being best practice for mobile UX.

There are so many factors that play a part in giving the user the optimal mobile interaction. You dumb it down to one option and it will fit some like a glove but be terrible for others. Also plenty of interaction innovation that is now widely used, has originated from UX/web designers, not the manufacturers.