According to William Dyer of the Department of Plant Sciences and Plant Pathology at Montana State University, U.S. farmers spend approximately 58% of the $11 billion a year in pesticide sales on herbicides - chemicals designed to kill pest plants, like weeds, as opposed to insecticides or fungicides. The large majority of herbicides on the market now were discovered using the cumbersome guess-and-check method. Scientists would spray chemical mixes on different plants and observe the results, slowly narrowing down the group through a series of trials to a few lead chemicals until one was determined to be the most effective.
This approach to herbicide development is dubbed the "chemical approach," meaning that scientists start with the chemical and see what eventually works. The chemical approach has worked surprisingly well and has been helped along recently by biotechnological advances such as quantitative structure-activity relationship analysis, which quantitatively relates chemical structure to a biological response and allows certain trends of biological activity to be discovered. However, new technologies may soon allow the old system to be updated.
For instance, the ability to sequence the genes of pest plants could lead to new targets for herbicides. Only a few enzymes are currently targeted by herbicides - those which are affected by the chemicals already developed. With a completely sequenced genome, additional enzymes revealed by the sequencing could be targeted with inhibiting chemicals specific to those enzymes. This approach, where the enzyme is targeted first and the chemical developed later, Dyer has called the "targeted approach," and it could lead to far more refined herbicides. It could also assist in the development of selective herbicides, where only the weed is targeted and the crop spared, since their enzymes differ.
Enzymes aren't the only possible targets for herbicides. Other genes, such as those that encode proteins essential for transducing signals or those that supply regulatory functions, could also be lethal to a plant were they turned off. Unfortunately, even a fully sequenced genome provides no clues at all to the function of each individual gene. The solution to this problem is to use the variety of new techniques for silencing genes, such as RNA interference, to turn off selected genes, identify those which greatly contribute to growth, determine their exact function, and find a way to inhibit them.
This targeted approach provides huge advantages over the chemical based approach. In addition to creating more targeted (and therefore, possibly safer) herbicides, the gene identification techniques may lead to a much more complete understanding of other processes which govern plant behavior. And it's definitely better than a one in 30,000 chance.