I spent the last year gathering and normalizing as many resources for Usenet messages and built a search engine over public Usenet discussions going back to 1981. Posting here because this seemed like the one group that would actually care.
The index currently holds (to date) roughly 980 million messages and is still growing. Sources are archival backups, data donations, and ongoing crawls of several thousand still-active public NNTP servers. Binary and yEnc-encoded content is stripped where possible to keep the index focused on text.
The index was built using Apache Solr 10, MariaDB 12 (rocksdb), Ubuntu
22.04.5 LTS & custom Python scripts, all on nvme disk.
This is public archival content, same legal basis as DejaNews and Google Groups before it. There's a removal process for anyone who wants their own posts taken down, and author contact info is masked by default.