One could also argue, that people, who use code from a repository without a license, don't want to make that fact publicly known. That kind of usage might be il

One could also argue, that people, who use code from a repository without a license, don't want to make that fact publicly known. That kind of usage might be illegal, so following or staring that repository might be similar to admitting a crime. Therefore, your metric could have a large selection bias.