Hey Yoav,

First, I think that the phenomenon of over-selling (in title, abstract, the way you refer to your results as ‘novel’ and so on), is part of wider trend in CS and academia in general.

Researchers have found correlation both with date of publishing (an increase in positive words over the years, specifically in medical papers) and with the competitiveness and productivity of the publishing university (across all disciplines).

Reference #1: http://www.nature.com/news/novel-amazing-innovative-positive-words-on-the-rise-in-science-papers-1.19024
Reference #2: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0010271

Also, I strongly agree that while the fast iteration model that arxiv allows has significant upsides, it also has major downsides, one of which in enabling (and perhaps encouraging) flag-planting.

These types of responses and discourse on arxiv-first papers, however, can help mitigate this effect, so thank you for that. It is a load that needs to be shared, of course, and so it puts more responsibility on use research students and researchers everywhere to read carefully and challenge such papers every once in a while (when they’re in our domain of expertise). Crowd-driven peer-review, in a sense.

Perhaps it should even be constructed in some way into arxiv, constructing a rating system for papers that incorporates not only originating university/research group, citations and peer-reviews (for papers that have them), but also on-site reviews, challenges and critiques, especially for papers before peer review.