Playing Hide and Seek in Second Life
Second Life is a big place. A VERY big place. You only need to consult the Second Life Grid Survey – Region Database (©Tyche Shepherd 2011) to feel a bit overwhelmed with its size. With over 31,500 Regions online, it’s not hard to get totally lost. And I don’t mean lost as in “can’t find my way home,” but lost as in “where’s waldo?” Lost in the crowd … Buried in the noise … One leaf in a forest of trees.
If you want to play Hide and Seek, Second Life is an ideal place to do it. The problem though is that most folks that are in business in Second Life want anything but to play games with their visibility. They want to be found, to be easily seen, and to be recognized for who they are and what they provide. The way they accomplish that is to set their Parcel or Store to “Show In Search.”
In Second Life, “Search” is handled by an impressive piece of technology called a GSA (short for “Google Search Appliance”). Like its similarly named big brother, the GSA crawls the entire Second Life Grid, indexing Parcels and their contents, storing all the information in its internal database. When someone uses Search to find something of interest, the GSA uses their request to look through its database and pick the most relevant results.
At least, that’s how it’s supposed to work …
The Secret Recipe of Search
When the current GSA first arrived in Second Life, it was attended by an avalanche of “issues”. Start up bugs creating the web pages from a Parcel’s contents, indexing issues with various words, biased results due to noisy data sources, etc. all combined to make the initial roll out very painful. However, once the Search Team dispatched with the basic technical issues, they began “adjusting” Search to try and provide a balanced and fair selection of results.
Search Engines use various measures and conditions to determine the “Rank” of a particular result. Collectively called “Ranking Factors”, their purpose and intent have been shaped by years of experience with searching big places (like the Internet) by companies like the creator of the GSA, Google. How those factors are combined, weighted and used to calculate the final rank is called the “Search Recipe”.
Unfortunately this is the point where the noble goal of providing a fair and balanced Search collided head on with the twin desires of “Proprietary Secret” and “Maintaining Control”. In Linden Lab’s goal of providing a useful and powerful search tool, they have forgotten and thus lost sight of the real purpose of In-World Search. By adjusting and reworking Search on a continual basis, they have turned it from a useful and beneficial service into a frustrating game of loss and confusion.
Stability is Life in Business
Turn on your television and change to your favorite news channel. Chances are you didn’t have to go hunting for it, in fact you probably knew the exact channel number. This is because the channel numbers are very stable. They don’t change from day to day, or even month to month. If channel numbers changed every week, chances are good you would give up trying to watch television altogether because the effort of finding your favorite programs would just be too annoying.
You’re probably asking yourself “Why would they want to change the channel numbers?” The answer is very simple. You see, stations that have low numbered channels get the most visibility. If your station is lucky enough to be in the first 40 channels then you are generally guaranteed to get a larger number of viewers. So if your cable company routinely shifts stations around to give them “equal time” in the lower numbered channels then it will be more “fair”. And we all know that fair is universally good, right?
Wrong. The purpose of television is to watch what you want, when you want to watch. It’s not a game of fairness. It’s a service meant to provide the expected programming when you select the desired channel number. Constantly shifting stations around the channel lineup would go against the goal of the service. However this is exactly what Linden Lab’s Search Team has done by continuing to adjust Search. They have taken the viable service that provided stable and expected results, and in the name of fairness, twisted it so it returns results that are neither relevant nor expected. In short, they have killed the one factor that matters most to any business … Stability.
Gaming, Spamming and Utilizing
In-World Search has been the target of a lot of angst and vitriol since it was upgraded last year. One of the most often targeted facets is a thing called “Spamming” or “Gaming” of Search. One faction argues that the use of specially named prims to gain visibility in Search is wrong, that it is against Linden Lab’s Terms of Service (TOS) and that it is a form of cheating. The other faction maintains that use of such methods is the only way to gain any visibility at all.
A lot of the passion involved in the argument centers on the definition of the term “Gaming”. Linden Lab has very little to say on the topic, however there is a single sentence in their Mainland Policies Wiki document under the “Policy Prohibiting the Misuse of Bots (scripted agents)” section as follows:
“Attempting to gain an unfair advantage in the Second Life search tools, including but not limited to the use of Bots or camping for that purpose, is a violation of this policy and (for non-mainland estates) of our Terms of Service.”
I suspect that the phrase “unfair advantage” is part of the reason that the Search Dev Team has gone to such extreme measures to assure the “fairness” of the results returned by In-World Search. However their efforts have had exactly the opposite effect. They have instead granted an “unfair advantage” to those that use named prims to gain better visibility for products and services they neither sell nor represent while completely removing (or demoting to completely invisible locations) those that were using named prims to engineer a proper and fair measure of visibility.
Is This Really Fair?
This point bears reiteration, so please do read this very carefully. There is a contingent of merchants in Second Life that have well known brands. They have quality products, they are dedicated to marketing, advertising and promoting their brand name and products, and they are willing to take almost any measure necessary to ensure their business succeeds. But they are NOT engaged in unfairly manipulating In-World Search to accomplish this goal. They are honest, decent businesspeople. They recognize the products they sell, the market they sell to, and they do not wish to even intrude into markets or harass people that have no interest in buying their stuff. And yet these are the very same people that have been slapped down, and slapped down hard, by this latest change in Search.
Late last year, the Search Dev Team implemented a rapid succession of changes that skewed the Search Results to give “unfair advantage” to owners of larger parcels. Jack Linden denied this of course, but everyone even peripherally involved in In-World Search knew it was true. When they finally (and quietly) turned the bias down some, they also adjusted the Ranking Factors to give an advantage to having large numbers of prims with names and descriptions that contained the chosen keywords. Their rationale appears to be that larger parcels can host a larger number of these (otherwise useless) prims and thus would once again dominate in Search. But this also resulted in a very unfair disadvantage to smaller parcels.
Birth of the Spim Prammer
I am deeply involved in studying and understanding In-World Search. After my rather distasteful encounter with Jack Linden, I realized the goals of their efforts. But since they had opened the door and seemed intent on giving unfair advantage to selected merchants, I decided it was time to bring the technology I’d been developing to the public market. I had a device in development that I had jokingly named the “Spim Prammer”. It was designed to rapidly rez a large number of prims, properly named and marked to Show In Search, and with tweaks to each to prevent the prims from triggering the spam detector developed by LL. It had minor success in the market, but I wasn’t advertising it and didn’t really mention it to but a few folks … yet it did begin to pick up steam as people found that it allowed them to regain proper rank in the market niches where they should rightfully rank.
The Last Change of 2010
Right at the end of 2010, they released an “Update” to Search. As with the previous updates, there was no notice given and no explanation as to what had changed. However, it had the immediate effect of knocking the top stores out of Search and promoting a completely random set of unrelated stores and parcels to the top of the ranks. My store was one knocked from a solid #1 position (with only about 10 objects marked to show in Search). Suddenly those places that had worked hard to maintain a good level of visibility were gone from Search. In their stead were unrelated parcels, places that had a totally random selection of named prims in an unfair attempt to gain top ranking for every keyword under the sun. For a brief time, their efforts were rewarded by the Search Dev Team’s change as they were vaulted to the top and their traffic numbers began growing by leaps and bounds. (Keep in mind their sales did not enjoy such great improvements as they were not really “relevant” locations and didn’t actually sell anything that people were truly looking for. But their only goal was to increase traffic anyway … and the change in Search did exactly that.)
After about a week of studying the “New Search” results, I hit on the solution … and built the changes into the Spim Prammer. I automatically sent updates to everyone that had purchased the prior version, notified my customers of the changes and how to use it to regain their proper visibility. Within a short period of time, merchants around Second Life began to realize that the only way to garner any visibility in Search was to “Obey the Rules” laid out by Sea Linden and her team. They could either manually reproduce the techniques embodied in the Spim Prammer’s operation, or they could purchase the Prammer itself. Once again a level of sanity was restored to In-World Stores and Search became a tool that they could again use to help support their business efforts.
At least it lasted for a while .. until the most recent destructive change on February 15th …
The Latest Tilt of the Table
For some unfathomable reason, stability in the market and of the services provided by Linden Lab appears to be something they strive mightily to destroy. On February 15th, Sea Linden let loose her latest brickbat to the Merchants of Second Life. With no notice, no explanation and no reason they released an update that immediately decimated sales and the results returned by the GSA. Since that day, myself and almost every other merchant that has an In-World store has been deeply engrossed in trying to understand how to regain some semblance of visibility. Unfortunately Sea has “Gamed Search” in such an obtuse and unfair manner that no one can actually compete anymore. Worse yet, just like the previous changes, the majority of “Winners” are those stores and parcels that are not actually relevant to the search. There are some resulting searches that, were it not that real people are losing real money daily, would be hilarious. For example:
- All Search, General and Moderate Maturity, Search Term is “condo rentals” – The first seven results do not even contain the word “condo”, it is only found in the 8th result listed. Exactly how are those seven properties “More Relevant”?
- All Search, General and Moderate Maturity, Search Term is “teleport” – The #1 and #3 results are an Art Studio and a Nude Beach. How exactly are these two places relevant to “Teleport”? (Hint: They have marked their teleporters to Show In Search to unfairly gain traffic.)
- All Search, General and Moderate Maturity, Search Term is “teleporter” – The #2 result is a “Language school, center and club”. They have also marked their teleporters to Show In Search to unfairly gain traffic.
These are just three searches that I have used over the past 2-1/2 weeks in an attempt to gain some understanding of how Sea Linden and her Search Dev Team have unfairly gamed Search. I have piles and piles of searches that all yield results that are completely nonsensical. These are searches that used to return viable results, searches that took shoppers to stores that sold or provided what they were looking for, and searches that people began to count on because they were stable and dependable. But then again, stable and useful results are NOT the goals of the Search Dev Team.
Conclusion … For Now
Sea Linden has rigged the game. While this may seem like it is her right to keep the playing field active so as to ensure a variety of results, this update has done anything but. Her changes, and moreover her disabling of the Boost value as meaningful data, the 12 hour delay before reindexing now imposed on Parcels that have made changes, and the fact that she has removed every single useful and comprehensible ranking factor from In-World Search, have all resulted in a steep decline in In-World sales all across the Grid. Her unfair and abusive manipulation of Search has driven people to the wrong parcels, destroyed any sense of competence in Search, and dealt massive hits to many many Merchants.
I have spent many hours trying to understand and compensate for her abuse of the service that she is supposed to provide, but so far I have been unable to make any sense of it at all. Several days ago I witnessed a parcel that removed eight objects from Search … objects that did NOT have the desired keyword in their name or description. By all sense this should have given them a slightly higher rank for that keyword, but did it? Nope, not at all. In fact it dropped them three places!
In-World Search is a Service, it is a TOOL meant to be used by Residents to find the things they want and need. Instead Sea Linden has turned it into an annoying, useless and unfairly biased waste of time and screen space.
In the last week, I have been approached by many people frustrated with the sudden and surprising downturn in their sales. Two of those (so far) have left Second Life for good, one of them abandoned two Sims and has instead chosen to simply suck money out of Second Life without offering any benefit or growth. (And to be honest, I totally understand and agree with their reasons for doing so.)
Hindering my efforts are the facts that Sea has extended re-indexing times to twelve hours AFTER a parcel’s objects change. If your parcel does not change then it is reindexed every 6 hours. But if you make any changes, your parcel will not be indexed for at least 12 hours (and sometimes as long as 24). Sea has removed the effect of the Boost Rating on Search Rank (as evidenced by the fact that for many searches, the parcels at the top all have a -1 or worse Boost while those with 0 Boost are demoted further down page one or even pushed back to the second or later pages.
These impediments are even further confounded by the fact that the program that turns a parcel’s object listing into an HTML page for the public to view is NOT the same program that generates the source for the GSA to index. These three processes (Public HTML, Private GSA-Only HTML and GSA Index) are “out of sync” with each other, thus causing them to update at odd times and often resulting in the GSA missing changes.
While you may be thinking this is all just sour grapes on my part, keep in mind that In-World Search is a SERVICE that is meant to provide Residents with rapid access to those places within Second Life that actually matter to them. But Sea Linden has done the exact opposite. She has purposely and destructively promoted parcels and Sims that have no relevance to the Resident using Search. She has destroyed the income of many good and honest Merchants in Second Life … and she has done so with out a single whisper of explanation or reasoning. In short, she has “Unfairly Gamed Search” so that it only returns the results she wants to see, and penalizes anyone else that might dare to try and compete in the business market of Second Life.
Once again this is a perfect example of how Linden Lab equals FAIL!