I think Rails and Django are great, but they offload a lot of design decisions to the community -- this is great when you want to write maintainable code with a diverse team, but not necessarily the optimal setup for gaining the most useful skills for the future. If you want to be ahead of the curve, try using smaller and more focused tools. Even though an individual piece of the puzzle may occasionally need to be replaced, you will at least know that you hand selected the best tool for your use case, and won't have to shoe-horn another approach on top of it.
Really, the choice of the tool comes down to a lot more than just "what's best today". But some common criteria are:
Language - are you comfortable with the language used in the tool?
Conventions - are you comfortable with the conventions used in the tool?
Community - what kind of community is around the tool? Is it supported sufficiently by helpful peers?
Fit for consumption - does the tool really do the job well? In other words, are you using a tool with the wrong kinds of strengths? (Programs written in C are fast, but writing raw C doesn't make sense for most web apps)
Team - Does your team use a particular tool? Are you collaborating with people, and if so, do they have an opinion about the language and conventions used?
Beyond this, it really has to do with aesthetics. Most any tool out there will "do the job", and with reasonable performance. Of course there are plenty that have significant gaps (compare a Wordpress load overhead to static files served by Jekyll, and you'll see an example of this), but at the end of the day, people utilize and patch and hack and ... get the job done.
What looks exciting to you? What are you going to put your heart and soul into learning? What is enjoyable? What works? Choose that.