[Craig] FERGUSON: No, the experiment always works. There’s no such thing as an experiment that doesn’t work. There are only results, but results may vary. Here’s what I learned:
Experiments can fail if they are executed or planned improperly. If both the control and the experimental group are given sugar pills, for example, or the equipment fails in a shower of sparks, the experiment has provided no evidence by which one can update. It is a small quibble, and probably not what the quote meant to illustrate (I’m guessing that the experiment provided evidence which downgraded the probability of the hypothesis), but something to note nonetheless: experiments are not magic knowledge-providers.
Experiments can fail if they are executed or planned improperly. If both the control and the experimental group are given sugar pills, for example, or the equipment fails in a shower of sparks, the experiment has provided no evidence by which one can update.
I think Ferguson would call those “results,” and from those you would have learned about performing experiments, not about the original hypothesis you were interested in.
If anything, I think a really failed experiment is one that makes you think you’ve learned something that is in fact wrong, which is the result of flaws in the experiment that you never become aware of.
I think Ferguson would call those “results,” and from those you would have learned about performing experiments, not about the original hypothesis you were interested in.
Ferguson’s proposed new language is a downgrade. Being unable to identify something as a failure when the outcome sucks is fatalism and not particularly useful.
I tend to disagree.. I have done some things which I thought was experimenting with but did not come up with any clear conclusion after the experiment and analysis. On rewriting the thesis it turned out there were a lot more implicit assumptions inside the hypothesis that I was not aware of. I think it was a badly designed experiment and it was rather unproductive in retrospective analysis. I suppose one could argue that it brought to light the implicit assumptions and that was a useful result. Somehow(not sure how or why) I find that a low standard to consider something an experiment.
Experiments can fail if they are executed or planned improperly. If both the control and the experimental group are given sugar pills, for example, or the equipment fails in a shower of sparks, the experiment has provided no evidence by which one can update. It is a small quibble, and probably not what the quote meant to illustrate (I’m guessing that the experiment provided evidence which downgraded the probability of the hypothesis), but something to note nonetheless: experiments are not magic knowledge-providers.
I think Ferguson would call those “results,” and from those you would have learned about performing experiments, not about the original hypothesis you were interested in.
If anything, I think a really failed experiment is one that makes you think you’ve learned something that is in fact wrong, which is the result of flaws in the experiment that you never become aware of.
Ferguson’s proposed new language is a downgrade. Being unable to identify something as a failure when the outcome sucks is fatalism and not particularly useful.
— Robert Morris, quoted in Brian Snow’s “We Need Assurance!”
I tend to disagree.. I have done some things which I thought was experimenting with but did not come up with any clear conclusion after the experiment and analysis. On rewriting the thesis it turned out there were a lot more implicit assumptions inside the hypothesis that I was not aware of. I think it was a badly designed experiment and it was rather unproductive in retrospective analysis. I suppose one could argue that it brought to light the implicit assumptions and that was a useful result. Somehow(not sure how or why) I find that a low standard to consider something an experiment.
An experiment is supposed to teach you the truth. If you run the experiment badly and, say, get a false positive, then the experiment failed.