i think they mean they’ll provide direct access to data hosted by "third party"s (torrents?), without the captchas and throttling/rate limiting present when normally using the annas archive website

they’re asking for text extraction and dedup in exchange for providing datasets. at least publicly they claim this whole project is aimed at data preservation and wide access… they’re mostly aggregating/collecting data from other shadow libraries and even if they have malicious(?) intent, i’d say they’re a net positive since their code and datas are mostly(?) open sourced.