Well, the list of tools is at least as useful as the most useful tool on the list, and probably reflects the overall usefulness of its category. Which means that if we go enough levels farther, the master list of lists [of lists...] would reflect the overall usefulness of all software on GitHub.
Also maybe GitHub needs a "list of stuff" category, so these lists of things don't muddle search results and conversely can be searched for easily. Curated lists of links are generally quite useful and putting it all in a public vcs with a pull request infrastructure makes it easy to contribute.
I definitely see a usefullness in these lists, I just feel sad that they get a lot more attention than actual code repositiories.
Indeed, Github should find a solution for this. I didn't care much about Github stars but recently I noticed they're being required in various scenarios, for example some CDN require a certain number of Github stars to include a library [1]
I don't think Github stars should be considered as anything more than built-in bookmarking. Personally, I only use them that way. With this view, it's obvious why lists get more stars than the projects listed.
If they do any tempering, expel them.
This requires a rich person to pay them, or a community to motivate them, which gives the super-power to an agreeable, but fair BDFL.