Well, I had hoped someone else might have saved me.
Let me present the two cases. In the first, I have drawn the resonance structures for nitrobenzene and benzonitrile. The arrow shows where one might expect electrophilic attack to occur from. It is the most electron rich by virtue of it not being electron poor. Compared to benzene, one might anticipate these compounds to be less reactive than benzene.
If I had to predict at which position substitution would occur, I would have guessed meta to the nitro as nitro is a more potent electron withdrawing group. I would expect the two oxygen atoms could draw electron density to them more strongly than a single nitrogen. I believe you can find this is generally true from the Hammett sigma values as well.
Now, I have also drawn the more traditional analysis of the intermediate resonance structures. In this analysis, we must decide that a positive charge adjacent to the positive charge of a nitro group is the most destabilizing. That would make the nitration meta to the CN an unfavorable position.
A similar analysis for nitration at the alternate position puts the charge adjacent to the CN group. Here, a formal charge is not present, simply an sp carbon. Since an sp carbon is the most electron withdrawing, it too would no be favored. However, if one were to predict, one would predict this is where substitution should occur.
As I stated earlier, if the substitutents conflict in direction, I would anticipate a mixture to form. If as in this case, the major product (50.1 v 49.4) is in fact the unexpected ortho to the nitro, I cannot provide a satisfactory explanation.
Again, I prefer the upper analysis as it does not require drawing as many resonance structures, it does not require one to recognize one of the resonance structures as destabilizing or a greater preference, and it does not suggest that they actually form. I believe the rate determining step is the actual attack. If attack did occur (as it does) at positions that give the lesser contributing resonance structures, they still go to product. I also find it easier to predict whether a compound will react faster or slower than benzene from the charge present in the resonance contributors.