You make good points @astrid. For what it’s worth it appears that PubPub is a non-profit project, and the organisation behind it is a partnership between MIT Press and the MIT Media Lab: About · Knowledge Futures Group
That said, longevity is a difficult question no matter whose servers the papers are archived on. The advantage of PDFs is that they require very little effort to host and easily fit with institutional repositories. On top of that they integrate with existing research workflows and bibliography management software, can be read offline, etc. It becomes easier to have several mirrors of the proceedings which don’t need special maintenance by us.
I believe the plan is already to archive PDFs either on Zenodo or nime.org, or both. But in the current version of PubPub, the quality of the PDF seems like an afterthought – the text is unreasonably large, the formatting is nowhere near as professional as LaTeX, the title page doesn’t look great. Fundamentally the PDF just doesn’t look like many people expect a research paper to look, and that hurts researchers (especially early career) who face pressures to justify their work to people not familiar with NIME.
There is a secondary issue to make sure the archived PDF (not just the HTML paper) is properly archived by Google Scholar, though scraping correct metadata is a challenge even with the existing template.
I think these issues are fixable, depending on the priorities of the PubPub team, but in the short term we have to carefully consider the tradeoffs.
Incidentally, the tradeoffs might be different with something like a NIME journal, where we don’t already have existing infrastructure and we might want a deeper integration of companion materials as part of the identity of the journal. I could imagine a scenario in which we used PubPub or a similar platform for a journal even as the conference proceedings returned to the old templates.