Sometimes the news is slow, and sometimes it comes in an avalanche. This past week or so has seen a whole lot of new and interesting things happen in relation to Second Life. Meeroos, a timeline for Mesh, The Great Facebook Avatar Deletion (with an added helping of I Told You So). All of these things are absolutely worth columns of their own (and fortunately, I get one of those every week). But this week I'm going to talk about the fact that Linden Lab claims that they've fixed the in-world search engine by completely rebuilding it, and that finally (in theory), it's going to become something other than "busted." Yes. I'm going to say it *AGAIN*. It's currently BUSTED, and believe me, "busted" is the nicest thing I can say about it.
As for why I'm going with this topic first, before any of the others? Nothing makes a site with the word "search" in the title happier than an article about...search. Right? Right. So let's take a little journey and have a look at the past, present and potential future of this much needed fix.
The Ghost of Searches Past
Image Credit: MetaVerso
Second Life has its own in-world search engine. That is to say, while you are logged into SL itself, there is a search engine which can be accessed within your viewer. This search is divided up into different categories in order to make searching (presumably) simpler. Prior to the release of viewer 2.x, search worked pretty well. By comparison to what we've been living with for the past 14 months, it worked amazingly well. Businesses within Second Life were able, via various methods to apply SEO tactics and the search engine would respond with a fair degree of predictability. Mainly, what worked so well was the results you got were relevant to what you were looking for.
This is not to say it was perfect. It wasn't perfect, but no search engine is going to get it right every time. The main flaw I personally found with it was that it had no fuzzy component. If you misspelled something you were screwed. This was often problematic in some types of search queries, particularly avatar names. Because usernames are required to be unique, there are an enormous amount of "creative" spellings in SL, and blowing the spelling of a name could easily lead to your never actually getting a relevant or accurate result. But when searching for things, places, or events, it worked quite well.
It worked particularly well as we look at it from where we stand now, because 14 months ago, in (roughly) April of 2010, everything changed. From there, things got ugly very, very quickly.
When It All Broke Down
When viewer 2.x was released in February of 2010, the way search was handled within the viewer changed. Though 1.x viewers still used the old search method, 2.x viewers used a completely different system, based on the GSA 6 system (the previous system was a GSA 5). Aside from the fact that many things in viewer 2 seemed to change simply for the sake of doing so, one of the reasons search changed was that the old search would use sim resources (whatever sim you were on) to route certain types of queries (mostly traffic data)to the search servers. It was decided that in order to reduce the load on these resources a new method was needed that avoided this problem (presumably to reduce lag and other sim strain issues). The new search system (GSA 6) bypassed sims entirely. In addition, Linden Lab had its eye on pushing various parts of the platform (including profiles, another cause of consternation- don't get me started) to the web, rather than being handled exclusively within the viewer. So reinventing search to reflect that reality was something they were eager to acheive. This new incarnation of in-world search used Google Search Appliance 6 in order to make it function.
The problem, of course, is that when they changed it around, they began to tinker with it. Incessantly. At some point in the spring of 2010 (March-April, depending on whom you ask) search broke.
Now, I don't mean broke in terms of "If you put in a query you'd get no results." I mean "if you put in a query, you very often got barely relevant results.", or "if you put in a query, you would get nonsensical results." This was especially problematic for those engaged in commerce within Second Life, as on what seemed like a weekly basis, search would change again, and not only make no sense in terms of what results went where, but became absolutely and utterly impossible to apply SEO. While the Lab claimed they were "optimizing" or "seeking to prevent gaming the system", what really happened was that real people lost an absolutely enormous amount of real money as businesses that had traditionally thrived got buried in search in favor of things like yard sales. The howling from the merchant community in specific and the SL resident community in general was incessant and (afaiac) completely justified, as the Lab used search like a tinkertoy, and every time they messed with it, the immediate costs were borne by those who were pushing the SL economy. If you were engaged in SL commerce, at some point (or a whole lot of points) you wound up talking to other merchants in the same boat- all either screaming, collectively scratching their heads, or both- often at the same time.
This problem became so widespread and severe that quite a number of merchants could not weather the hit. They either closed up shop, either having lost so money that maintaining their business beecame unsustainable, or so frustrated and angry with the entire thing that they simply refused to keep trying to do the equivalent of Charlie Brown trying to kick a football.
Unfortunately, when things like this happen, with sure predictability, the Lab attempts to dodge the issue. They refuse for quite some time (ages, if you're the one losing money) to even acknowledge there's a problem, insisting instead that it's YOU who has the issue, even avoiding plain evidence of a before and after comparison. When that changes is when the loss of revenue eventually trickles down to the Lab itself-- when people can no longer afford tier (sim rental) fees, which cost up to $295USD monthly each and "tier down", because they cannot afford the cost. That costs the Lab real money, and it's generally only then that they start to (grudgingly) acknowledge that there's actually something wrong, and the residents aren't simply making it all up.
In addition to all of this is the observation that in the past year the Lab really hasn't cared all that much about in-world search. In fact, it could be said that it was to their advantage that they simply let search knowingly remain busted. This is because the Lab's focus has been rather agggressively on SL Marketplace, at the direct expense of in-world shopping, a practice that has been cited by a number of people (self included, right here in this very column) as being short sighted. The Lab makes a commission cut on every Marketplace transaction, something that does not happen when an item is bought in-world. Eventually the comparison becomes "all those Marketplace micropayments" vs. "How much is being lost in tier costs from people who say screw it, and tier down."
But for 14 months now, in-world search has remained a serious, serious problem for those whose businesses rely on it.
Into The Present
On May 25, it was announced that (finally) something was to be done about in-world search. It even included a Torley Linden video (seen below). Every time they get Torley to try to sell something, you know the Lab *really* wants you to buy into it.
However, despite all the "ooh shiny" there's some other things in that announcement. Namely this bit (emphasis mine):
"New search will soon be available to you in the official SL Viewer and we will not be implementing it for the 1.23 Viewer. To be clear, you can still use the 1.23 Viewer, but search functionality will be impaired once new search is released into general availability, after the test period."
Which means, ladies and gentlemen, that as I said months ago, Linden Lab is going to break the current version 1.x search. Though third party viewers based on 1.x code will have the option to implement this new search system into their viewers (Linden Lab says they plan on putting the new search into 1.23, but no timeline at all has been established on their doing so, which I tend to view as "never"), most third party viewers either have, or are trying to adopt a 2.x codebase, in order to keep current with new SL technologies. For those people who are still firmly resisting moving to a v2.x based viewer, some hard decisions may be on the horizon.
On Friday afternoon, a meeting with third party developers and the Lab took place, to discuss this issue. Here's some of the highlights that may be of interest:
- The new search system is out and live in a LL project viewer. (it should be noted that You do not need the new viewer to test the new search function).
- The new search system is going to completely replace the one currently in use. (Hallelujah).
- Third Party Viewer devs will get 60 days notice before viewer 1.x search effectively breaks.
- An API will be made available for viewer developers to get the data and present it without having to use the web page presentation.
- The new search system is planned for live release to Second Life residents on June 13.
- The old (current) search system will be going away in July.
- The new search system is not based on GSA at all, but rather on SOLR/Lucene, which will cost the Lab much less money to maintain. This search system is the one that has been installed into the new SL Marketplace, so this move will mean that both in-world and marketplace searches are operating under the same system.
The Ghost Of Searches Yet To Come
Of course, we won't know how well this new system really works until it's been in place a while, as results and changes will not be instantaneous. However it's very important for the future of Second Life commerce that not only this new system work, but that it work with minimal problems. The past year has been incredibly hard on SL merchants, and should this be another disaster even more than have already thrown in the towel will have no choice but to do so.
For some merchants, it's already too little, too late. Having been burned so badly by the search fiasco in the past year, they will not return. Some simply stopped doing business in SL. Some took their business elsewhere, like inworldz, or Avination. Some did a combination of both, or simply expanded into other grids to attempt to stop the hemmoraging of money stemming from the loss of sales in Second Life.
It's also important that this work strictly from the Lab's point of view. This is a moment at which they could potentially lure people back into the virtual world to shore up losses from the past 18 months or so. Saving search, improving it, and creating a situation by which more money were pumped into the in-world economy can only be a good thing for Linden Lab on all fronts, not the least of which from a public relations standpoint.
We will know soon enough. The merchant community in SL, already frustrated and skeptical of the Lab's ability to frankly so much as water a houseplant, will take some convincing. But if they can pull it off, the community will be grateful, as the influx of income tends to blur the edges around a lot of old wounds.
The Original Search Marketing Event is Back!
SES Denver (Oct 16) offers an intense day of learning all the critical aspects of search engine optimization (SEO) and paid search advertising (PPC). The mission of SES remains the same as it did from the start - to help you master being found on search engines. Early Bird rates available through Sept 12. Register today!