Thanks.

I was also thinking about saving gossip but there are already people doing this and I am not sure it makes sense to duplicate the effort (https://github.com/lnresearch/topology) especially since it takes huge amounts of disk space. From this collection I was just able to download just a few days of samples but I suppose they have everything.

What would be cool to do with such data is then some ranking on nodes. Based on first NodeAnnouncement you could find out the exact age of a node. And then you could also calculate historical "uptime".