Notebooks mix scientific text and code. The code part of a notebook may be called as any Python module from another notebook. This awesome feature makes Zanadu not only a distributed collaborative library but also a structured architecture of interconnected research modules.
New innovative algorithms have to be benchmarked with respect to the state of art. Just call the state of art algorithms when they are available in Zanadu and run any kind of test/memory/speed/accuracy study. Of course, comparing algorithms will always remain a complicated matter. Yet Zanadu allows for a direct access to those algorithms, so that benchmarking has never been so easy.