How The Globe Works

Sure, The Globe looks pretty simple: rez a sphere, slap an Earth texture on it, and boom you're done, right? Uh, OK! But a little more happens behind the scenes to get everything else working. At a general level, The Globe uses:

  • Off-world database services to manage questions and updates
  • Explore Mode uses both the GeoNames geographical database service and Wikipedia to identify locations, and then Wikipedia and Wikimedia Commons for descriptions and location-based images

The Trivia Game

Questions

By default, The Globe fetches questions from an off-world database. Each time The Globe moves to a new category, it retrieves a new set of questions from the database. This way, instead of having to manage thousands of questions, The Globe only deals with five questions at a time. Downside: The Globe can go offline if there are Internet problems between Second Life and the databases, or the databases are otherwise inaccessible.

Within each category, the database chooses questions randomly. Some categories are quite large (like World Capitals) and might contain hundreds of questions. Other categories are smaller (a category on recent Olympic Games might only have ten questions), so players might see repeated questions if the category comes up more than once.

But that should take a while! When the Globe asks for a new questions, the database chooses question categories semi-randomly. Basically, categories get chosen randomly unless a particular Globe has used a particular category recently. This is intended to keep categories from repeating too quickly, so you shouldn't finish a set of questions on World Capitals, then have The Globe announce an exciting new category: World Capitals! (But it might pick something like Most Populous Capitals or Least Populous Capitals or Former Capitals or another category with capitals!) Categories can repeat, but it will take a while. When categories do repeat, odds are pretty good most of the questions will be different.

Questions can be in more than one category. For instance, "Who can click closest to Beijing?" might appear in the World Capitals category, as well as Most Populous Cities. A question about the Great Pyramid at Giza might appear in a Wonders of the World category, as well as Famous Landmarks. You get the idea. However, most questions are just in one category.

Pins & Distances

Players often ask how The Globe calculates the distances between an answer and a player's pin. The Globe uses a minor variation on the Haversine formula, which gives the shortest ("great circle") distance between two points on a perfect sphere. The Globe uses a minor variation because:

  • the geographic data behind the questions is based on a very slightly ellipsoidal Earth (in real life, the Earth bulges a little around the equator)
  • the accuracy of click data on a Second Life sphere varies with location

The latter consideration is by far the most important. With a pure Haversine formula the worst-case accuracy for a click on The Globe was ± 40km. With some minor tweaks, The Globe is accurate to within just under 10km in the worst case scenario (and considerably more accurate most of the time). Not good enough to land a plane by, but good enough for a trivia game.

Explore

What Happens When You Click

Explore Mode has more going on behind the scenes than The Globe's trivia game. Every time a user clicks a location, these things happen:

  1. The Globe works out the latitude and longitude of the clicked location and sends it to its offworld service
  2. The offworld service asks the GeoNames geographic database for information about that latitude and longitude. That generally means "the most significant named location reasonably close to the clicked location."
  3. The service then asks Wikipedia if there's an entry about that named location.
    • If Wikipedia has an entry, the servce tries to create a summary paragraph, and then asks the WikiMedia Commons if there are any images for that location. If images exist, the service randomly selects from images it thinks might be most visually interesting. The number of images selected depends on the image viewer being used with The Globe: the viewer built into The Globe can handle up to four images.
    • If there's no entry, the service "backs out" to the next-most significant named area and tries again. For instance, if you click on a small village in Siberia and no Wikipedia entry is available for that village, the service will try to get information about its administrative district. If nothing is available for that, the service will try to get information about the village's oblast (province or state), and finally about Russia itself.
  4. The service bundles up that information and sends it back to The Globe, which displays the name, description information, and images (if any) as best it can.

What Happens When You Ask For A Location?

When you ask for a location in Explore Mode ("Where is Sugarloaf Mountain?"), the process is similar to clicking a location, except GeoNames isn't involved. The Globe tried to figure out the location name based on the chat text, and then asks Wikipedia if it has an entry for that place.

The Globe will try some variations on the name if Wikipedia doesn't have an entry for the specified item, but The Globe will not try to "back out" to the next most-significant named area. When you ask for a specific thing, either The Globe can figure it out quickly or it can't. That's all.

Assuming an entry is found, the process is exactly the same as a clicked point: The Globe asks Wikipedia if it has a description and images for the item, tries to pick some interesting pictures, and displays the results.

Why Do Clicked Results Vary So Much?

Basically, more information is available for some places than others.

If you click at a major urban area—particularly an older urban area like London, Mumbai, or Beijing—the density of significant, named places close to the point where you clicked is much higher than, say, clicking the middle of the Greenland ice sheet. (Although if you get lucky, you might just click the Summit Camp there!)

That said, the GeoNames serivce and Wikipedia's ideas of what's significant doesn't always match what you or I might think is significant. Trying to click London to get a picture of Big Ben or The Eye can be incredibly frustrating because The City of London itself is really quite small, and there are so many other significant locations around it. On the other hand, trying to click a specific, isolated location (like a ghost town in the American west) can be very difficult because the geographic database and Wikipedia may not represent it well enough for the location to even be clickable.

So, if you know the name of what you're looking for, just ask for it directly in local chat! "Show me Big Ben!" "Where is the London Eye?" "Move to the City of London." Those should all work! (For a ghost town…um…how about "Show me Bodie California?")