29
votes

I would like to know some examples of GIS projects that have gone wrong.

i.e. Bad decisions would/were made, based upon the GIS output from that project.

I am not looking at naming and shaming anyone, so if it is a company you have worked for, please keep any names out of it, and disguise the story where necessary. I am sure there must be quite a few major public examples?

e.g. A disaster mgmt project, gone wrong due to bad or out-of-date data.

I admit, it is a fairly subjective question with no right answer. Therefore, I will let the votes do the talking, and if someone could convert to c-wiki if need be.

Personally, I will get great benefit from this, and I will be passing the examples on when teaching GIS.

0

11 Answers 11

32
votes

Not really a full-blown GIS projects, but interesting example of situation when spatial aspects were not handled correctly.

Economist's 'North Korea missile threat' first looked like:

wrong

Whereas in reality should look like:

correct

Readers quickly pointed mistake and Economist admitted it and posted correction.

(via Spatial Analysis blog)

Edit: Some more info about the problem from ArcUser Online.

5
  • 5
    dam round earth always getting in the way, why couldn't it just be flat :P
    – Nathan W
    Apr 1, 2011 at 13:42
  • 1
    What is the projection they used in the correction? What made it more appropriate than a mercator?
    – Nathanus
    Apr 1, 2011 at 15:20
  • 4
    @Nathan W - maybe the Flat Earth Society were GIS developers fed up of dealing with projections? Apr 1, 2011 at 16:58
  • 1
    @Nathanus: some more info on ESRI blog blogs.esri.com/Support/blogs/mappingcenter/archive/2009/07/15/…
    – radek
    Apr 1, 2011 at 17:31
  • 1
    I saw at the ESRI Fed UC that they're adding support for this sort of analysis (basing buffers on projected data) -- cool stuff.
    – Roy
    Apr 16, 2012 at 11:21
11
votes

Here is one more: Eurocrats forgetting about Wales on the cover of Eurostat yearbook 2004

enter image description here

Reactions from BBC, Telegraph, and Mail.

(via Spatial Analysis blog; image from GIT NEWS)

9
votes

A great question to ask over beers (you'd get some fun answers for sure).

My greatest failures (yea plural unfortunately) usually revolve around scope creep. More is never better, words to live by.

9
votes

Should't Cristopher Columbus be mentioned here?

He thought he was on the other side of the globe when he arrived in America. No doubt it was because of bad map data.

edit: maybe at least not only bad data according to whubers comment, more like common architect technique to get their dream project come true.

/Nicklas

4
  • 9
    Actually, he had reasonably good theoretical models and fairly good data about the size and shape of the earth. In order to get funding he had to make the project look practicable and so he used lower-bound estimates of the earth's size to guess the distance westward from Portugal to the western Pacific ocean. (He was low by about 12,000 Km, as I recall. :-) This practice of optimistically pressing the bounds of uncertainty in data might look familiar to many readers...
    – whuber
    Apr 1, 2011 at 14:37
  • 1
    good point. the same concept is used in every bigger building project as well I think. so there should be some doubt in my answer, I will edit :-) Apr 1, 2011 at 14:47
  • 1
    Without having a reliable clock, he couldn't accurately compute longitude. Moral to story: time is an important factor in any project. Feb 28, 2012 at 19:16
  • 2
    @whuber - fantastic - "optimistically pressing the bounds of uncertainty in data" is the most succinct phrase I've ever heard to describe the sorts of external pressure one gets during analysis (or imposes internally, I guess).
    – Simbamangu
    Oct 29, 2012 at 18:18
7
votes

Jeez, I could mention a few. My main issue about GIS projects, is that they are thought of, analysed, designed and built as IT projects with GIS elements, rather than GIS projects. The importance of data and data management always appears to be dropped and, to be frank, they are quite a few people out there who don't know enough about GIS to be involved in their planning.

I once knew of a project, canned after 6 years of being way, way, way over budget, whose Lead BA's depiction of the Great Britain was a 100,000km spherical coordinate system.

If a project has an element of a GIS, or is entirely a GIS, it has to be planned, from the Get GO as a GIS. To may times it is built as an IS with some maps tacked on, and they always fail miserably.

5
votes

Many of the failures I've seen have been a result of Big Design Up Front, which then led to a sunk cost dilemma.

4
votes

I have one. Workflow for a GIS Project. If you don't get this right you will spend up huge amount of time and money for an organization. Example is designing a data model for collecting utility based networks. As a project manager you have to spend bulk of your time preparing a data model for GIS Project. There are organizations that use their own data model (usually old). ESRI's has so many industry-specific data models that one can utilize in their project rather than starting from scratch.

4
votes

Spatial Sustain points to one more example : Map Satire Gets The Economist in Trouble.

The Economist published a piece on Scotland and the cover of the issue looked like:

According to the blog:

The Economist, uses a map to poke fun at a post-independence Scotland, with such new location names as “Skintland” for the country, and “Loanlands” and “Edinborrow.”

...

Emotions are running high over the slight, with the First Minister even saying that the magazine will “rue the day” it poked fun at the Scots. The map is being called insulting, offensive, racist, and likely worse.

Result: 1595 comments so far (as for 16.04.12). And pretty heated discussions ;)

3
votes

Depending on what you call 'go bad', but my impression is that most maps which are made by gis-scientists at universities are so bad that nobody uses them (except, perhaps, for climate change models).

Most common error: overly optimistic interpolation. Instead of using a global mean for an area/stratum, values are interpolated using a complicated technique which hides the fact that it is not more precise or even less precise than using a average/standard deviation.

See eg: http://www.springerlink.com/content/qq5h67635g4l4417/ or a lengthy discussion paper here: http://dx.doi.org/10.1016/S0016-7061(97)00072-4

Second (and related) most common error: thinking that the kriging error represents the error on your prediction.

3
votes

Good examples are to review to US-DOI/BLM projects called ALMRS or NILS. Both of these projects were terminated after 10's of millions of dollars (or more by some estimates) investment to improve accuracy to national land records.

Both were found to have decreased accuracy and in some cases exhibited gross waste.

In the end, it was found that commercial data sources, built using the BLMS own data were much more accurate and up to date than what the new BLM systems were doing.

So much was this a failure that the BLMs own offices were contracting to commercial sources as there needs were not being met by the internally developed systems. Comprehensive requirements were not being looked at for their own needs.

3
votes

Here is one more (quite surprising) example of Google causing quite a lot of stir in Japan by including 1858 (sic!) woodblock maps of Tokyo (Edo) in Google Earth.

Although map itself is gorgeous:

enter image description here

(interactive source)

it pinpoints locations of Buraku villages that used to be inhabited by 'untouchables'. And even after so many years there were fears that prejudice and discrimination might still be an issue.

I love Japan even more now! ;]

(more info on the issue here and here. ).

1
  • 1
    Very similar to the practice of red-lining here in the US.
    – Andy W
    Feb 28, 2012 at 18:20

Not the answer you're looking for? Browse other questions tagged or ask your own question.