Microsoft Lost a City Because They Used Wikipedia Data (theregister.co.uk)

"Microsoft can't tell North from South on Bing Maps," joked The Register, reporting that Microsoft's site had "misplaced Melbourne, the four-million-inhabitant capital of the Australian State of Victoria." Long-time Slashdot reader RockDoctor writes: Though they're trying to minimise it, the recent relocation of Melbourne Australia to the ocean east of Japan in Microsoft's flagship mapping application is blamed on someone having flipped a sign in the latitude given for the city's Wikipedia page. Which may or may not be true. But the simple stupidity of using a globally-editable data source for feeding a mapping and navigation system is ... "awesome" is (for once) an appropriate word.

Well, it's Bing, so at least no-one was actually using it.
"Bing's not alone in finding Australia hard to navigate," reports The Register. "In 2012 police warned not to use Apple Maps as it directed those seeking the rural Victorian town of Mildura into the middle of a desert."

Comments

Re:Not totally true

[dAzED1]

many people actually let google "know" who they are during their surfing, so if they google for the porn their google account has a record of it. Instead of doing the easier/better bit of opening a "private" tab or whatnot, they subject themselves to bing. That would be my guess for such a thing being true, if it is.

It's hard to believe.

[hey!]

The amount of data you need assemble a global navigation system is enormous. You don't hire some intern to transcribe data out of Wikipedia, you license it from companies like Tele Atlas.

Now for geographic place names you'd turn to sources like the USGS GNIS system for the US, whatever the local equivalent of GNIS is, or for places that don't have that datasets like GNIS the DoD's Defense Mapping Agency.

It can't possibly be that Bing gets their place/position data mainly from Wikipedia. The only thing I can think is that they did some kind of union of all the geographic name sources they could find in order to maximize the chance of getting a hit on a place name search, and somehow screwed up prioritizing the most reliable sources first.

Re:It's hard to believe.

[AthanasiusKircher]

The source for "we use Wikipedia" also said they just use it for metadata on locations and the actual location API didn't get the location wrong just the search engine subject result.

The thing is -- this is still disturbing on many levels. Repeat after me: Wikipedia is NOT a STABLE source of reliable information. Wikipedia is NOT a STABLE source of reliable information. The very idea that ANYONE is using Wikipedia for ANY application where accuracy or reliability may be desired is disturbing.

And now someone's going to trot out a reference about how Wikipedia is "more accurate" than Britannica or whatever. Maybe it is, at any given moment. And it's certainly more exhaustive on many topics now. But one significant difference between Wikipedia and more traditional sources is that Wikipedia can be edited by anyone, at any time -- and that can mean improvements, but it can also mean introducing errors, either accidental of deliberate (vandalism).

Paper Britannica, whatever its flaws are, doesn't spontaneously generate new typos every time I open it.

And a few years back we could all laugh this off while still praising the great things Wikipedia was doing. No more. Wikipedia's active editors are shrinking all the time, but the vandals show no sign of shrinking... if anything, they seem to be growing and becoming smarter to evade the bots that will revert obvious vandalism (e.g., random insertions of profanity). And that doesn't count random, well-meaning mistakes that people introduce or deliberate attempts to slant entries in a particular way.

And if you think you haven't seen vandalism, you probably don't realize how subtle it can be. I still remember a few years back when I discovered an odd century error in a history article -- something was listed as 19XX when it obviously should have been 17XX. Then I noticed another one in the article. I checked the edit history and found a vandal who had been going through and changing random DIGITS in DATES for history articles. The ones I saw were rather obvious. But in other articles he had just changed a year or decade, in which case few people were likely to notice the discrepancy. These edits had been live for weeks... nobody had noticed them.

Such vandalism might stay in entries for years. Media sources now use Wikipedia frequently and even academics sometimes take info from it without verifying. (Particularly for something that would seem obvious, like a basic date for a common event.) Now what happens when those academics start putting the wrong dates in books because they read it on Wikipedia? Before, when I saw a different date in a history book, I might even assume it might be because there was some scholarly debate over when an event happened exactly... now it could just be arbitrary randomness introduced by the vandals.

The ONLY thing we have going for us to protect Wikipedia is that MOST of these vandals can't help themselves. They don't have the initiative to build up a reputation of a few good edits before embarking on a mission of anarchy, nor do they have the discipline to refrain from just being a troll. The date-editing vandal I mentioned above was eventually caught and his edits reverted, but ONLY because he finally posted "X is a dirty slut!" or something on a prominent female historical figure's article, leading someone to check out what this guy had been doing. But as I said, his edits stood for weeks without anyone noticing. Do we really want to rely on the trolls outing themselves as our main safety net??

Go to more obscure articles, and lots of bad stuff can happen. Just a week ago I was reading an article on American history, and I encountered a few sentences that seemed quite surprising and unlikely. They seemed to be propaganda based on current political stuff going on in the U.S., but it was framed as though it was a part of a historical event. I looked some of this up, and couldn't find any source for this -- I'm reasonably cer

Re:It's hard to believe.

[Anonymous Coward]

As far as I'm concerned, the worst problem isn't the vandals, they're annoying, but while Wikipedia isn't 100% reliable on all data at any given moment, as a whole it's reasonably reliable and if an article is cited properly (not all are, but there's enough to make it very useful) data can be verified before using it to do anything where the info is truly critical. M$ not doing that is not surprising, they're pants-on-head retarded.

No, the worst problem on Wikipedia is the deletionists. These people are the worst, most insidious kind of vandals. They get their jollies by finding an interesting article and declaring it "not encyclopedic" and removing it, or finding article content that someone has researched, cited, and written, and destroying it. They make editing very frustrating, because you never know when you'll spend a few hours writing a new section of an article based on some new research and having one of these assholes trash it.

I had this happen, I had come across some information on a possible new experimental use for an existing drug. I researched, wrote (and thoroughly cited) a section in that drug's article, very carefully specifying that the new use was experimental. Again, this was backed by multiple "reliable source" citations as specified in the Wikipedia "rules".

Less than a day later, my work was gone. The deletionist had declared my information "not accepted medical practice" and removed it. Well, duh, it wasn't "accepted medical practice" - it was experimental, and I had carefully noted that.

I had another article, a list article of what was essentially trivia that I had been periodically adding to, vandalized by another deletionist, because not every entry in the list was likely to result in a full article. The article was essentially blanked, and became utterly useless, and then the vandal nominated it for deletion. I managed to prevent that, but it took a lot of time and work to get the list back into reasonable shape after that.

A third article I got involved on was a stub article on a small town. The deletionist vandal wanted the town deleted. I spent a few hours on that, researching the town and citing it, and managed to get the deletion stopped - but I don't have infinite time to watch the deletion queue to save articles from these vandals.

It's extremely discouraging when you spend hours working on something only to have one of these deletionists destroy it. I consider them far worse than the obvious vandals, because the destruction of information is their stated goal, and it takes far less work to destroy than to create, and these vandals have seemingly unlimited time to destroy.

Re:Not totally true

[ArtemaOne]

Thank you. Bing sounds smarter than "tricking" Google by switching to a private browser coming from the exact same IP address.

Amazing they get so much right!

[Theovon]

I realize Melbourne, Australia is a big deal, and it seems like with a city that large, Microsoft and Apple and others could afford to hire one person whose job is to make sure they get stuff right.

But at the same time I find it amazing that they don’t have more mistakes. The navicable roads across the whole world are vast, and living in an imperfect world, there’s alway going to be some probability and degree or error in everything we do. Getting Melbourne’s location wrong because Microsoft may have copied Wikipedia is funny. But when it comes down to it, for all the things they could have wrong, this mistake constitutes a SINGLE BIT error. Yeah, it’s a super big deal bit, but in terms of raw information content, you have to be surprised that they don’t suffer from single-bit errors all the time in less significant but noticable ways.

Also, given what we all SHOULD know about science, we should understand that every model of anything is going to be correct only within certain statistical bounds. Yes, that the universe was smaller in the past and has to have been dense enough to have undergone a phase change (cf. CMB), so the big bang as a whole is essentially settled. However, there are details we don’t have filled in yet, so whenever someone comes out with some new alternative to inflation, we look at it with a critical eye. We should be doing the same when it comes to these electronic gadgets we use. There are many different failure modes. When we become so trusting and dependent on them that we can’t recover from their failure, then we’ve got a problem. They’re never going to be perfect. Moreover, different services will implement different algorithms that will give us different results. When navigating somewhere, you need to use your brain to decide which route is best, not just trust what the routing algorithm says. Moreover, local knowledge always trumps an algorithm whose knowledge of traffic patterns and back roads is extremely limited.

Let me give you an example. Let’s say I’m a little to the east of Binghamton University on Vestal Parkway. If I ask either Google or Apple Maps where the nearest gas station is, they BOTH give me a location on the opposite (north) side of the river in Johnson City. Why? Because they use cartesian distance. As the crow flies, that gas station is the closest, but to get there, I would have to back-track to the west to 201, take it north to Riverside Drive, and then back-track to the east. Either that or try to drive across the river. A much FASTER gas station to get to from there (although with only a slightly shorter total driving distance) is actually in Binghamton, to the east, on the same (south) side of the river, where there are no turns or traffic lights in the way. In other words, these routing algorithms are stupid about rivers and other common traffic phenomena. And of course none of these have a way to consider the fact that I actually live in Vestal and am likely to want find a gas station between where I am and my house. Sure, they’ll list multiple gas stations, and I can choose the right one, but this is an example of needing to use my brain to make the decision, rather than relying blindly on software.