I agree that a shorter staking period would likely make STEEM somewhat more attractive to investors. On the other hand, the security of having a long staking period should not be discounted, especially given that a) active keys are handled frequently, b) there is currently no hardware wallet support, and c) I'm not aware of good options for notification services to help warn users of an unexpected powerdown. Also, taking the experience of EOS into account, we should be concerned about the possibility of exchanges voting with user stake.
With these things in mind, currently I would caution against entertaining anything less than a 4-week powerdown. Even then, I would personally want to see a) some good notification services, and b) a powerdown schedule that is back-end heavy...e.g. 4 increasing weekly powerdowns of 10%, 20%, 30%, and 40%. That way users still get 100% in 4 weeks instead of the current 13 weeks, but the new risks are mitigated fairly substantially, and without adding much complexity, compared to 4 equal weekly powerdowns of 25%.
Lastly, I see the benefits of having some kind of dynamic staking scheme with variable APR, which could enable staking options both shorter and longer than 4 weeks (where longer staking periods would come with more voting power). But obviously this is more complicated and would take more time to come up with the most sensible (and simplest) possible scheme, and to code/test it. Not to mention, I would not favor enabling any staking periods shorter than 4 weeks without first having hardware support in place.
P.S. I get that items such as notifications and hardware support are not core level changes.