Interesting. Got this:

Nonsensical. You can't accurately measure a player's strength by giving only 30 seconds to tactical puzzles.
When I took it seriously, it gave an estimate of 1580 points, when I deliberately made nonsense moves in one second on each puzzles, it gave an estimate of 1300.
Baaah, don't be silly. Of course you can measure a chess players skill with accuracy, you just don't understand what a bell curve or confidence interval is ;)
Seriously you don't get statistics. It isn't accurate. It isn't not accurate. It is 95% confident you are in the range stated. Moreover you are giving anomolous data to a statistical model. It assumes you are demonstrating skill and not fooling it. All your 1 second guesses are outliers and should not be used for predictions.
The whole system is almost certainly using machine learning algorithms which do not respond well to bad input. If you give it nonsese data instead of showing it your skill of course it will not do a good job. However your nonsense data is likely being added to a databse and used for other predictions. So infact you are probably making it less accurate with every dumb move you put it. Now get back to school hyyyahhh 
If someone has a 13xx rating here maybe their 1-second responses really do accurately reflect how they think and what their rating range is. I know that most GMs' first guess would still put them at GM levels.
Apparently it estimates me at 1712. Kind of matches up to my chess.com ratings or at least in the right ballpark of where they have been within the last 6-12 months. Not convinced by it though. When I am in a game I have the 'feel' for the game through having played all the moves to get to whatever position I am in. I'm certainly not a deep calculator of variations, but just dropping into a position with 30s to size it up and play a move doesn't feel to me like it matches up to the situation I am in when I play a serious game (which I've only done in one tournament in the last 15 years!).
My approx elo is 1933. Sure. I lose to 1700's all the time. You can't estimate my elo from only 30 tactics. You need a lot more data, like actual gameplay. Just because you can solve a mate in one doesn't mean your positional play is good. I read positional chess books all the time, and yet I still have a hard time grasping some positions that look equal from the Silman Thinking Technique, but in just 20 moves somehow white/black gets an advatange (with best play!) and wins...
I know my rating. If I stop losing to underrated youth players, my rating will match my skill level.
OTB gives you a real rating. These estimates, on the other hand, http://chessskill.blogspot.com/2009/05/rating-estimation.html
I think the idea of this test is not to measure your ability to solve puzzles, but to compare your answers with what players with known rating answered and based on that it is guessing your rating.
thanks, really cool site. i didn't answer the ratings questions at the end as i worried they somehow impacted the ratings they gave you. i'm a 1300-ish player on chess.com where i play the majority of my chess. this website gave me a rating of 1340, so pretty good estimate. i get that it was a timed tactics test and some might not think that's a good gauge, but the fact is you either know tactics/patterns or you don't, so someone know nothing about tactics probably isn't going to score over 1000 in a timed test like this.
ok, so an update. i sent this link to a guy in my office who has played a few games but has done very little reading or studying. he scored a 1350. i think he has a natural gift with the game but he has no formal training or ability to know fork/mating patterns, yet scored high on the quiz. this is particularly troubling for me as i've played daily for 4 years, done literally thousands of puzzles and scored a bit more than my playing average. happy for him, hope he continues playing, but am now ever more aware of how mediocre i am!
http://www.elometer.net