I think the SV influence is probably much stronger than those numbers show. For one thing, what are the numbers like when adjusted for posting frequency? Five users who each post once a month don't balance out even one of the people who post a dozen comments a day. Also, people who have very strong connections with SV companies as employers, investors, or even customers are likely to adopt some of those attitudes even if they live elsewhere (like me). I'm pretty sure a real sentiment analysis would show a different result than a mere user survey.

The numbers vary depending on what we measure but they just don't vary that much. For example, when you look at total submissions and comments, the SV portion goes up to 10.9%. The highest SV number I've seen is actually for total page views (14.4%), which may suggest that SV users are reading more than posting, relative to other demographics.

When I read your comment I can't help but hear it as taking a step away from the numbers and back into preconceptions (i.e. real numbers would show what you already know). One thing I've noticed from years of doing this is that people's preconceptions about HN are amazingly strong and often not open to change. Sorry—I know that sounds condescending; in case it helps, I don't think I'm any different.

> users who post submissions and comments

That's a far cry from weighting by comment instead of by user, not to mention the other factors I mentioned. Maybe it would help if you could provide a link where people could look at the raw data (properly anonymized of course) and do their own analysis instead of relying on yours.

> people's preconceptions about HN are amazingly strong

Indeed. As you point out, nobody's exempt. Of course, there are ways around that. The proof's in the pudding. There are probably many people here who could do a real sentiment analysis and compare the result to other sites. I'm as sure as you are of what that would show, even though we make different predictions.

I don't understand what you mean by weighting by comment?

The corpus of HN comments is public and available to anyone who wants to do a sentiment analysis on them. I've not seen any compelling-enough examples of sentiment analysis to want to do it.