First, there are some very serious problems with those "benchmarks" --
they aren't really comparing apples to apples at all. There's some
interesting things to dig into with that work, but stating
categorically that this "proves" that Django is "slow" is... a
stretch.
Second, raw speed isn't everything. Scalability, ease of use,
suitability, and many other factors play into the decisions we make
about technology. If we only cared about raw speed, we'd all still be
writing assembly.
Finally, of course we care about performance, and we're constantly
monitoring and improving Django's speed (see
https://github.com/jacobian/djangobench, for example). If you'd like
to help, there's plenty of low-hanging fruit, feel free to dive in.
But let's not pay *too* much attention to a badly-designed study with
unclear goals and intentions.
Jacob