New Scientist on GM Crops

Chuck Grimes cgrimes at
Sun Mar 26 17:51:47 PST 2000

So just how are scientists supposed to show that bacteria are unlikely to pick up and activate bla? How would a properly designed experiment to show this look?

Ken Hanly ------------

Well, the way you have worded this, makes it impossible, which was my point. You can't show what doesn't happen. You can only show what does happen.

In any event, instead of nit-picking I'll assume you have worded the question the way you may have intended it to be asked.

The general approach to a counter-positive involves developing sufficient positive information to surround what you want to demonstrate is unlikely. In other words, you show all the positive results the do happen to eliminate the alternatives. Then you arrange a series of experiments that demonstrates these positives in turn. These should constitute as many or ideally all of the alternative events.

Then under this logic, you have demonstrated everything the does happen, and the event that doesn't happen, is argued to be the last and unlikely other alternative.

It is usually empirically impossible to cover an exhaustive list of alternatives, so this approach is rarely used in anything as messy as biology.

Your example of the 100psi tested hose normally used at 40psi can be made an example of how to surround an unlikely event with most of the likely events.

You can't just run the hose for T hours at 100psi and declare it safe to use at 40psi. If it doesn't break, then you have demonstrated nothing.

The hose has to be tested to destruction to establish its likely breaking points.

You have to set-up times and pressures over various ranges to establish a series of demonstrated failure points. These points have to show a relation between time and pressure. Then you can deduce the time to failure at 40psi. Notice that in all of this, we are discussing what is likely and what does happen, not what doesn't.

Chuck Grimes

More information about the lbo-talk mailing list