Traffic changes

There are two new blogs on traffic changes and I don't see how they change anything at all:

https://blogs.secondlife.com/community/land/blog/2009/08/21/changes-to-the-traffic-scoring

https://blogs.secondlife.com/community/technology/blog/2009/08/21/search-rank-and-traffic

Traffic has never been a huge factor in search all, it remains the dominant factor in places. How long an avatar remains on a parcel is important for social places but for stores, especially well designed ones, you've always missed out by not keeping people on your parcel and nothing in these changes alters that issue.

Liana's blog is very explicit about that. "In the short term, it will have little or no effect." The other changes aren't happening until later, and LL won't be sharing the details.

The traffic formula they claim they are going to (1 point per avatar-minute), is no different than the result I have been getting for years by observing how many people are there on average times 1440 minutes per day.

If they had been doing some complicated formula involving 5 minute blocks and percentage of your login time, well they were just wasting cycles calculating it.

The easy way is just have the simulator count how many people are in a given parcel once a minute, and increment a table of parcels in the region by the right amount. Then once a day upload the totals to the home office (global parcel database for search). The simulator already knows which parcel you are in, cause it has to send that info to you to display on top of the viewer window.

As far as search ranking algorithms for Search/all, they already have made silent changes twice at least this year that I know of. Early this year they started counting keyword spam *against* your search rank, and more recently started listing wiki pages.

They've never shared the details on how traffic factors in search all and their understanding of how traffic scores are calculated seem wide of the mark. As I said, nothing seems to have changed.

The current scheme really is pretty transparent, the factors are enumerated in the knowledge base and it's not hard to approximate the effect of each with simple experimentation. This last part is the flaw that LL needs to eliminate.

What will change is that the number we see in About Land won't be used for searches the same way, if at all, the future. Lists of other metrics will be added. Some anti-spam weighting a la Google is in the works. These new metrics, unlike the old ones, will not be shared.