It is fiendishly hard to get cooperation from the people you'd really like to draft for surveys, and the traditional methods of doing so are losing their effectiveness and defensibility. Newer methods that rely on an all-volunteer army of online respondents are increasingly coming under fire because the participants don't mirror the general population. As a result, the entire opinion industry will need to be remade. It's getting harder for pollsters to put together a good sample to survey, which leads to a greater risk of inaccuracy and significantly increases costs. Stanford political scientist Douglas Rivers foresees the death of telephone polling. As an alternative, pollsters are looking toward online panels of volunteer respondents. Web surveys, which are already used by virtually all consumer researchers, cost about one-tenth as much as phone surveys, and it's easier to query a volunteer panel than to question a random sample of the populace. But online panels bring their own serious problems: While the general population is becoming less and less eager to respond to surveys, the most active web panelists are, strangely, more eager. Some of these "professional respondents," as pollsters call them, volunteer for 15 to 20 surveys a month. They're also demographically unrepresentative of the general population--more educated and more likely to be white, and more likely to be either young or old (but not middle-aged). The American Association for Public Opinion Research, a standards-setting body made up of nearly 2,000 survey professionals from private and government organizations, takes a dim view of web panels that are drawn from volunteers. Stuck between traditional polling, which is getting more difficult and expensive, and cheaper surveying of unrepresentative panels, the field is due for a radical rethink. At least one thing is clear: In their quest to probe the public mind, survey firms will make greater use of the vast amounts of digital data that people wittingly or unwittingly provide about themselves.