Traffic in Second Life – how does it work?

The question of just what traffic numbers mean and precisely how the numbers are generated has come up many times and many replies point to a rather out of date article in the Knowledgebase. Here’s that entire article:

Quote: What is Traffic (Dwell)?This document was originally written on 09-24-2004 by Haney Linden, and updated on 03-04-2005 by Jeska Linden. Please note some of this info is out-of-date and kept for historic purposes only—specifically, we no longer have monthly US$ Developers Incentives, but the general principles of traffic still apply.

What is Traffic ? (Dwell)
Linden introduced “traffic” or “dwell” as a way to reward residents who create popular locations. Each night a pool of $L is paid out to landowners in proportion to the number of avatars visiting and the amount of time spent on any of their parcels. The dwell that any one avatar triggers in a day is divided up proportionally to the time spent on any given parcel.
The total dwell generated by an avatar is the same whether they spend 5 minutes or 24 hours in SL on a particular day.

· An avatar must spend at least 5 minutes on a parcel for any dwell to register. Landowners who own several parcels next to each other may benefit from joining them in case some avatars spend less than 5 minutes on any parcel.

· Each month, Linden offers a $US reward to the 2% of landowners who receive the most dwell – in proportion to the dwell their properties received. In August of 2004, US$2,308 was distributed to 36 landowners.

· Dwell on group land also counts. The nightly $L that results from dwell on group-owned land is distributed equally to all members. However, the dwell “points” that count toward the monthly US$ awards are distributed to the members who contributed land allocations to group land, in proportion to their contribution. This offers a better chance for someone in the group to receive some of the US$ awards. We started tracking this September 1st so the September US$ awards will reflect group land.

· Both visible and invisible Private Islands receive dwell.

· The nightly L$ pool and the monthly US$ pool increase regularly, based on the growth of Second Life.

· Dwell also determines which “Popular Places” are selected each night. You can see these by clicking on the Find button and selecting the “Popular Places” tab. They also appear on the Map as a “thumbs up” icon.

· The dwell of all avatars count, including the 7-day trial accounts.

Group Dwell
Some more information about Dwell and Groups:

Group members who contribute land allocations get credit towards these awards in proportion to the amount of land allocation they’ve contributed. The daily $L that results from traffic on group land is paid to all group members equally.

Private Islands also receive traffic and in most situations it goes to the owner of the island. The exception is if the Private lsland is set to a group, then the group members will equally share the daily L$ dwell. However, when calculating the monthly reward, all traffic goes to the owner. Private Island owners should make sure that the same group does not own land on the mainland, otherwise the traffic won’t be counted correctly. Also, if a Private Island is transferred then the traffic for that month won’t count towards the monthly award.

Another piece of text that gets mentioned when the question comes up is said to be found in the F1 help:

Quote:How Traffic Works.

Imagine that every Resident has one point of traffic they can bestow each day. For their presence somewhere to count, they must be at that location for at least five minutes. So if a Resident logs in for five minutes and stays on a parcel of land for that time before then logging out, that parcel earns one point – because they spent all of their in-world time there.

Now lets say the next day the same Resident logs in and spends one hour in one place, and one hour in another place, then logs out. Each parcel they spend time on will get half of the traffic, so each will earn 0.5 points. So even though the Resident spent far more time in each location – they earn less dwell than the place she spent five minutes at the previous day.

Finally, on the third day – they log in and spend eight hours in one place, and two hours somewhere different. The first venue will get 0.8 points, and the latter will get 0.2 point of traffic.

So – the traffic generated by a Resident has nothing to do with the length of time they spend somewhere. It is based on the proportion of their in-world time they spend somewhere.

Nite Zelmanov brings up these points:

Quote:Originally Posted by Nite Zelmanov
These posts all point to a KB article that doesn’t explain the traffic number at all. It says how you give out traffic, but not how the actual parcel traffic is calculated for display. Does each day show the previous day’s total? Is it cumulative in some way? Averaged? Who knows?!?

Kelly Linden says:

Quote:In theory traffic works like this: each person has X traffic to give, independent of how long they are online. That X is divided among all parcels they spent time on according to how much of the time they were online was spent there. Very short times on parcels are discarded (traveling over a parcel doesn’t count)
…
It is a 437 line perl script.

The explanation that an avatar gets so many points that get distributed to each parcel in which the avatar spends over some minimum amount in proportion to the percent of time they spend on each parcel is not real complicated, but it isn’t complete, and some folks don’t believe that this is actually the case, and offer alternative explanations of how traffic is calculated. Here is Kitty Barnett’s post of that nature:

Quote:Originally Posted by Kitty Barnett
I thought I had explained what the result was.

I started thinking the knowledge base article was how traffic actually worked, the only thing I was personally curious about was how many traffic “points” an avie got assigned in a day so I logged my alt on and let it idle for about 15 minutes and logged it off. The next day the traffic read 17 which wouldn’t account for places being able to get 60-80k+ worth of traffic.

Next was letting my alt idle for much longer than the original 15 minutes which according to what the Lindens keep telling, shouldn’t affect the generated traffic in any way as long as it’s the only place it visited that day. The next day just showed the amount of minutes I left it idle.

Next was letting it idle for 5 minutes there on the parcel and then idle for just under an hour somewhere else which if the article were true should give a negliable rating since it spent 10 times more time on the second parcel than it did the first. The result just reflected the number of minutes.

Idling along with my alt added twice as much which definitely contradicted the article since I’d be logged in for the better part of a day and only spent 10 minutes idling/testing so whatever I added should have been close to nothing, not the equal amount my alt did.

Take a look at what’s on popular places at different times throughout the day, and make your own average of how many people are on the sim during the day and then multiply by 60 minutes and 24 hours and you’ll get a number that’s on the same order to what their traffic reads the next day.

I honestly prefer the old way, it puts greater value on social gatherings since whoever attends those isn’t likely to idle online 24 hours a day and their traffic would be more valueable; with the way it seems to be right now, it doesn’t matter if you attract a zombie or a live person, they’re worth the exact same in terms of traffic.

Is it possible the traffic calculation method has been changed to something along the lines Kitty suggests and the support and documentation staff haven’t been advised of this change?

Nite’s questions “how the actual parcel traffic is calculated for display. Does each day show the previous day’s total? Is it cumulative in some way? Averaged?” bring up points that the explanation of how an avatars points get split up don’t cover at all.

I’ve posted essentially this same post in the Linden Answers forum;  I wonder what if any answer it will get.

I appeal to anyone who has performed empirical tests of how traffic works to reply here with a description of their experimental procedure and their raw data for analysis by other readers. You can post your conclusions but the data is the part that really matters.

Perhaps the confusion over the matter could be reduced if Linden Lab would make public the 437 lines of perl code, or whatever piece of source code is actually currently responsible for generating traffic numbers. I’m sure the that with over two million residents to look it over someone would figure it out.