PubSub Sandbox

Over the past few days, you may have noticed a few odd pieces of LinkRank data flowing through the system. There has been some backend issues that we are working diligently to fix. In fact, once we deal with these issues, LinkRanks will probably work alot better than ever.

We know how important correct statistics are, especially in the blogging community. Our users are very important to us and we continue to strive to provide the best data possible.

Thanks for your patience while we work on the issues.

Steven Cohen
PubSub Concepts, Inc
scohen@pubsub.com

The little rabbit that runs around the Web and brings back the LinkCount information for us failed to show up for work today (maybe because we’re paying him peanuts instead of carrots?)

Seriously, one of of databases that stays up to midnight to run LinkCounts failed us. That has been fixed and we ran the statistics again. You might see some inconsistencies for February 16th.

Sorry about that.

Note: No rabbits are harmed during the running of LinkCounts.

-Steven Cohen
scohen@pubsub.com

We’ve put together this FAQ on LinkRanks based on some of the feedback recieved over the past month.

---

Q: I publish to my blog everyday. Your calculations of the number of entries and outlinks for my blog is incorrect. What’s going on?

A: Every day, PubSub attempts to access the feed for your blog in order to count outlinks, inlinks and new entries. There are times when we can’t get to your feed due to our service “timing-out” on your server. If we can’t access your feed during a certain amount of time, a “time-out” occurs and we move on to the next feed. The next time we are able to fetch your feed, we will gather the data since the last time we read it. This is where the very large number for entries and outlinks comes in. Also, note that if a site has multiple feeds (i.e. RSS + Atom + RDF) we sometimes double count articles. We’re working on methods to reduce such overlap.

---

Q: Does this affect my LinkRank?

A: No. Linkrank is not based on the content that you post, but the number and quality of the inlinks that you get.

---

Q: So-and-so linked to me, but it didn’t show up on my LinkRank. Why?

A: The most likely reason for us not picking up an inlink is that the linking blog doesn’t have a feed or that the feed contains “summarized” entries. PubSub only reads the feed of the site.

---

Q: Why does it take so long for my new blog to get a LinkRank?

A: This is best addressed in conjunction with the FAQ on the submission page. It may take a few days for your blog to get a LinkRank. Using the same criteria as when your site is “live”, you will get a LinkRank the first time your blog gets linked to. This, of course, could take a few days once your content gets into the blogosphere and people locate and link to your blog.

---

Q: How is my LinkRank calculated?

A: If we told you, we’d have to kill you. Actually, we don’t provide the algorithm for LinkRanks for obvious reasons. If we disclose the details of our algorithm, spammers and sploggers would game the system. In order to provide you with the best possible “game-free” rankings, we don’t fully document the calculations. We regret very much that this is necessary.

---

Q: Why does my LinkRank fluctuate so much?

A: Your daily LinkRank fluctuates widely as it is inevitable, unless you have a very popular blog, that the number and quality of inlinks to your blog fluctuates on a daily basis. To get a more “stable” view of your LinkRank, take a look at the 15, 30, and 60 day average LinkRanks that we provide. Since these LinkRanks are calculated over a longer period of time, they fluctuate much less than the daily numbers.

---

Q: What does LinkRanks have to do with Community Lists?

A: We provide the LinkRank for every blog (if it has one) in our Community Lists, however, it is important to understand that blogs are put on the list as the result of a decision by the list editor that the blog is particularly worthy of attention. While the position of a blog on a particular Community List is determined by LinkRank, the decision of whether or not the blog is listed is determined by human editors, not by machines.

Today, we opened up a 30 day average on the Community Lists. A 30 day average of LinkRank provides a more stable view of influence over time. Result? The placement of the blogs will not jump around as much and one can get a better view of the top blogs on a more consistent basis.

For example, compare the daily ranks for the PR Community List…

…to the new 30 day feature

We hope that you’ll find this new feature useful.

-Steven Cohen

If you are one of those SiteStats junkies (if you have to ask, you probably aren’t one) who must know right away if your blog has risen or fallen in the rankings for the previously day, you’ve probably seen a marked improvement as to when your data is delivered. Our engineers have been working night and day to bring your rankings to you (Did you know that you can have them sent to your aggregator? Just click on the Atom button after running your URL.) as soon as humanly possible.

Also, since the Community Lists use LinkRanks, these too are updated just past midnight for the previous day. I can just picture the PR community refreshing their page every 5 seconds after the stroke of 11:59PM to see how they fared. If there is one community that loves the lists, it’s those in the PR industry. And we love that.

Hopefully, by the end of the week, I will be able to announce a new feature for the lists…

Steven M. Cohen
scohen [at] pubsub [dot] com

We are honored that Larry Chase has chosen PubSub Linkranks for one of his top sites for marketers for 2005. Here is what he has to say:

“This year, Internet marketers finally realized the impact of link popularity on search engine optimization. What this site can do is update you on a daily basis of new sites that have started linking to you. LinkRanks is PubSub’s method of measuring the strength, persistence and vitality of links appearing in the sources that PubSub monitors. It can tell you how influential a particular site is on a particular day and over time. With LinkRanks, users can estimate the reach and influence of a website based on the sites that link to it, which enables one to measure the effectiveness and impact of a link popularity campaign. In essence, if you score a link to your site from CNN.com or thousands of other major sources, PubSub will tell you what it’s worth. For example, at the time of writing three new sites were pointing to WDFM today; looking back over the past 30 days, our home page had a high of new daily links numbering 10. Not only that, LinkRanks tells you who they are. Thereafter, you can examine how important they are by seeing who links to them. Check out the home page to see who the winners and losers are in terms of daily incoming links.”

Thanks Larry.

Back on March 31, Bob Wyman announced the initial rollout of the LinkCounts page. We are pleased to announce that we have completed the first round of improvements to our weblog metrics package. These improvements include some changes in our processing as well as some new features, and put us in a better position to handle more sites and feeds as we rapidly approach monitoring over 10,000,000 weblogs.

We’ve reprocessed all of our raw data back to February 24th, 2005 and have replaced the older LinkCounts statistics with the freshly calculated values. We now have the ability to track the number of new entries per feed. We have also worked to break out more blog services and photo repositories into individual sites. For instance, in our “old” data, spaces.msn.com was consistently one of the top linked sites. Now we break out the individual blogs on Spaces into unique sites.

We have also continued to provide a certain level of filtering of the links that we count: things like tracking gifs, XML namespace links, emoticons, etc. Also, you can now view the Inlink and Outlink history of any site that we know about. And we are not done yet, we are continuing to refine our processing techniques and to add additional functionality. (And by the way, we haven’t forgotten about LinkRanks).

It’s been a very interesting experience watching the evolution of the Top 100 Sites as we have refined our ability to identify unique sites. Today, the majority of the top sites could be classified as quiz sites, dating sites, and photo repositories. The other fun aspect of this work has been to watch the growth in the number of feeds that we monitor. In Bob Wyman’s post a little over over a month ago, he noted the following statistics from the LinkCounts page:

Top 100 Domains By In-Links for March 31st, 2005

• 98,893 sites posted new material yesterday.
• They represent 1% of the 9,214,750 sites we know about.
• Yesterday 36,978 sites (37%) linked to other sites.
• There were 568,341 links created yesterday.

As I type this, the numbers now look like:

Top 100 Sites By Inlinking Sites for May 3rd, 2005

• 320,891 sites created 1,081,308 new blog entries
• 108,659 (34% of those with new entries) created 1,498,042 outlinks to 158,085 other sites.
• 6,566 sites (2% of those with new entries) had both inlinks and outlinks.
• 332,733 syndication feeds had new entries.
• 6% of the 5,362,137 recently active feeds monitored by PubSub had new entries.
• 3% of the 9,876,809 feeds monitored by PubSub had new entries.

Buried in the numbers is the fact that we added 622,059 feeds in 34 days! Thats an average of 19,472 per day.


It’s all happening at the LinkCounts page.

If you have any ideas for other statistics that we could calculate and present for you, let us know!