And please don’t get into predictive scoring. Go on fact and fact alone. We as users all understand that you can’t measure what’s not there. Time is against you for the immediate-need users and there’s not a person or company on earth that can affect time. Keep up the good work.

Melanie – You guys and gals are a responsive bunch, that’s for sure.

RSS – Fair point, but a blog that has 10K subscribers can make an argument for being of higher quality than one with 1K in a similar field. I agree with your thinking, but still believe there to be a place for it.

All other points, cool. Just again, please don’t get into predictive math. Just stick to the facts. You’ll be more credible for it.

Bravo to all. AideRSS is certainly helping me manage my feed reading (and making my blog look good, too.)

Dagnabbit, those new VPs are just too enthusiastic!

So yeah, Jim’s right – we specifically set out NOT to compare apples to oranges, because, as you note, what’s really the point of trying to compare pretty much anyone else to TechCrunch?

However, with the introduction of Thematic PostRank, if that’s what you want to do (or compare any group of personally relevant site “fruits”), have at it, that’s what it’s there for.

Our argument for not closely focusing on numbers of feed subscribers, for example (which we can glean), is that that metric falls into the same trap as most metrics out there — it only measures reach (quantity), not relevance (quality). So while there’s a good chance that if you write lots of good stuff if you have loads of RSS subscribers, it’s not as much of a guarantee as if what you write generates lots of engagement — that’s our argument, anyway.

Re. site traffic, if a publisher installs our widget, we can, in fact, track clicks and views. We also have a tracking code, which will be getting more implementations in the future.

Re. requiring “predictive” ranking to keep up with reader processing speeds (e.g. “wanna know it all by 9am), Jim’s got it. The only other side issue to that, as I see it, is that the immediacy need is very similar to the “professional vs. hobbyist” information need. So a professional marketer, for example, would probably want the social media coverage fire hose, whereas someone only beginning to explore what social media means probably needs/wants to start with the “greatest hits” to figure out what’s going on and how to get involved, and can take their time getting ramped up. People with a compelling event or pressing need require more time and content precision than casual users.

Re. the extension causing Google Reader slowness, we’ve heard this a few times, and while we’re constantly working to improve our system speeds as much as possible, there’s a user system side aspect we can’t control, so we’ll shortly be adding an on/off toggle to allow people the choice to speed things up if need be.

I totally agree on it being intermittently really enriching to be able to find the “hidden nuggets”, and alas, we don’t have a solid answer to that yet. But since it’s something I find valuable, I’ll keep pushing the devs with “what ifs” when I can.

The comment count issue is one we’re aware of and a fix is in progress. I’ll give you a holler shortly when it’s in place.

Whew! The comments are going to be longer than the blog post! Hope that sheds some light on things, though.

Any other questions or feedback, holler! (melanie@aiderss or @aiderss)

Great post…just a few clarifications.

1. As you say Feed-based PostRank normalizes engagement score relative to a feed’s hostorical engagement. This is what makes your 7.5 different from a TechCrunch 7.5. But, with the newly released Thematic PostRank (used in the latest GoogleReader Extension) the PostRank scores are normalized to each other.

Now you can use the API to send in urls to 50 posts on 50 different fees and get PostRanks normalized to that “theme”. Try it with posts form the VG top 25 list for example.

2. Regarding PostRank taking some time to return with meaningful results. We’re focused on continually reducing the latency of gathering engagement metrics that exist on the “Intenets” but there is a tougher problem when there isn’t any evidence of engagement…yet. We measure social engagement to a post – if there hasn’t been any then there is nothing to report. If we could predict the social engagement of a post with just the post itself that would be a neat trick but we’re currently not able to pull of that magic.

We’ve talked about developing a predictive model to initialize PostRanks/Engagement scores while a post is initially making its rounds. From a pure PostRank point of view we’d rather not muddy the waters by miking in “lies” to the otherwise defensible numbers. From a reader experience though I can see mixing in heuristics to get closer to the experience you’re looking for – but that would be outside the purview of the PostRank API.