Redmonk's Stephen O'Grady has a perceptive piece on product demos.
He makes the reasonable observation that, when trying to understand a new technology or new product, it's much easier (and fulfiling) if you can get your hands on it and play with it for yourself.
Quite rightly, he argues that slideware is far from convincing and managed demos (where you don't get to actually poke it yourself) are little better.
On one level, I have to agree with him: I spent several years travelling the world performing pilots, proofs of concepts and live demos of IBM's middleware products to potential customers. It was hard work, stressful, the hours were long and I lived with an ever-present fear that a mistake I'd made when developing the demo would cause the whole thing to fail when we presented it to the CIO.
But it was an extraordinarily exciting time and there is no better way to help people understand what you're offering them than letting them try it for themselves.
However, I think his call for vendors to "ship him the bits" could perhaps be worded a little better. Shipping working demos is the key. Just shipping the product and telling someone to install it and figure it out for themselves is probably not a good use of their time. Sending them a pre-configured scenario - with the ability for them to play with it and make changes - seems far more valuable to me.
Regardless, he has touched upon a key aspect of human nature: we tend to be far more visual than we would like to admit: a well-drawn diagram can obviate the need for three PowerPoint pages of text yet I - and many others - are guilty of ignoring this in almost every deck we produce.