I believe that the best science exhibits arise when you aim to intrigue and excite the visitor, rather than convey specific facts or concepts to them. In this respect, I think that classics like Thermal Camera and Slow Bubbles (illustrated) are wonderful exhibits. It takes a long time to teach someone something - more time than most visitors spend at a given exhibit or even at a whole science centre. But it's possible to inspire someone in a much shorter time. The effects of that inspirational experience, on that person's attitude to and interest in science, will probably be much deeper and longer-lasting than anything you try to 'teach' them with an exhibit.
That's not to say that it's a bad thing if exhibits have a direct educational function, and most if not all of the ideas I describe could be used to convey specific scientific concepts. But for me the first purpose of hands-on exhibits ideas is to astonish, intrigue, and delight.
Another plank of my philosophy is: make it real. Don't fake it. If you lead people to believe that something is happening, then it really should be happening.
Finally, get the ergonomics right.
Exhibit prototypes
These are ideas that have made it into working proof-of-concept models. Some of these items might work better as show demonstrations than as standalone exhibits.
Infinity machine
A set of gears, each of which turns more slowly than the previous one. The first gear spins merrily away, but the gear at the other end of the gear train might not make even one revolution in the life of the Universe...
Read more about the infinity machine.
Mystery lights
Two lights appear identical in colour, but objects change drastically in colour under one light compared to the other. As well as being amazing, this exhibit shows that we are all colour-blind.
Read more about the mystery lights.
The Bidwell effect
A Victorian visual illusion in which a red light appears blue for no obvious reason. An effect which has been unjustly neglected by science centres.
Read about the Bidwell effect
The more paradoxical scales
For most scales and seesaws, it matters where you put the weights. A small weight a long way from the pivot can balance a large weight close to the pivot. "Paradoxical" scales have been made where it doesn't matter where the weights are. I made some "More paradoxical" scales, where the large weight has to go further out than the small weight to balance it.
Read about the more paradoxical scales
Changing colours
You normally expect things to appear the same colour whether you are close up or standing back. The two lights in this prototype are different. Close up, they appear identical, but if you walk a few metres back, one of them looks pink and the other looks green! The effect is due to a little-known quirk of our colour vision.
Read about the changing colours prototype.