Saturday, January 24, 2004

Unknown unknowns

I spent much of today visiting with family - both my grandparents and my dad/step-mom. I just got the page proofs of my in-press journal article yesterday, and today my dad and I spent a fun hour or two chatting about the study and statistical analyses. Overall it was a good day.

My dad and I also discussed the recent press's handling of Donald Rumsfeld's "unknown unknowns" comment (see for example The Guardian). As my dad so eloquently said, "I don't exactly look for reasons to defend Mr. Rumsfeld from the press ..." but this recent series of jabs seems totally uncalled for. Here's the full Rumsfeld quote:

"Reports that say that something hasn't happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns - the ones we don't know we don't know." (from The Guardian)

I'm not saying the statement is a paragon of clear speech, however, the term "unknown unknowns" is not nearly as illogical as many reports seem to make it out to be. It's a term that was used in the space programs of the 60's and 70's (and possibly in other places before that, I don't know) to represent variables/items that engineers didn't know to pay attention to. Its abbreviation "unk-unks" is included in at least one list of military jargon, so I'm suspecting it's also used in the military.

Probably the easiest way to think about the term is that we can (as Rumsfeld explains) break our knowledge into three broad classes. First, there are things that are known, and thus we know that we know them; they are known knowns. For instance, we know that the high temperature in San Francisco, CA on Jan 23, 2004 was 53F.

Then there are things that we know exist, but that we don't specifically understand or know their state. These are known unknowns (in other words, variables that we don't know the value of). An astronomical example of this is "dark matter," which we know exists but do not know for certain what it is composed of. Another example is that we know San Francisco will have some high temperature tomorrow, but we don't currently know what that high temperature will be; it's a known unknown. If we wanted to, say, predict how far some insects could fly tomorrow we might want to pay attention to the high temperature tomorrow (our known unknown), and possibly try to model it.

Then, finally, there are elements of the world that we do not know exist. These elements are unknown unknowns. If you're trying to mathematically model something you could consider these as unknown variables; variables we don't even know about. The phrase thus serves as a useful tool to remind us that there are many things that are completely unknown (and thus we should attempt to search them out and/or plan for them).

A biological example of an unknown unknown I can think of is to go back in time, say, a hundred years and ask someone to model how far an insect could fly tomorrow in San Francisco. That someone would clearly know to consider the high temperature tomorrow (our known unknown), but they wouldn't know to ask what type of myosin/actin fibers are in the insect's muscles, since proteins hadn't even been discovered yet. In other words, the muscle fiber type was, a hundred years ago, an unknown unknown; nobody even knew to consider it when thinking about insect flight.

Another good example of an unknown unknown my dad thought of was the cellular damage caused by radiation. When early researchers were studying radioactive materials they knew certain things about the materials they were dealing with (their color, size, shape, etc), and they knew that there were certain things they didn't understand (e.g. what was being lost from an atom when it decayed). However, nobody knew that it was even possible that radiation would damage cellular components and cause disease; that was an unknown unknown (which lead to many untimely deaths).

From a teaching perspective I think the phrase may also be quite useful, as it conveniently describes the variables students don't know to pay attention to in their experiments because their world view doesn't include them (at least in relation to that specific experiment).

No comments: