Great idea! (assuming that ESPN and CNN can get it to work.) Payment always ought to be tied as closely to performance as possible. So why not tie valuation of ad time to the effectiveness of that time slot in connecting with consumers?
The possible results from an advertisement (or advertising campaign) fall on a spectrum of effectiveness that looks something like this:

An advertisement is slightly effective if it leads to brand recognition (consumers can remember having seen the brand when they are presented with the brand name or logo) or brand awareness (consumers can remember the brand when asked to name ads they've seen recently, or to name brands in a particular product category). But an ad is really effective if it generates word-of-mouth, if it gets people excited and talking about the brand with their friends; or, even better, if it starts a veritable ideavirus - one than spreads like wildfire from person to person simply because it is excellent, remarkable, and worth spreading.
Traditionally, advertising is valued based on CPM, or cost per thousand impressions. The number of impressions is the number of people who might potentially have seen the ad (assuming that they were paying attention, and hadn't muted the tv, gotten up to use the bathroom, or engaged in a side conversation while the commercial played). Thus, media "impressions" is quite a nebulous term - it gives no real indication of the effectiveness of the ad:

Setting prices for advertising space and time based on the number of people talking about the brand, rather than the number of "impressions," places value on the real results of the ad: its effectiveness.

I do hope that ESPN, CNN, and other media outlets will be able to find an accurate way to measure the amount of talking that an ad generates. Switching to a new pricing scale based on effectiveness will be a huge improvement to tv advertising, one that will benefit the media, advertisers, and viewers alike.