Unintended restrictions

Edmonton, Toronto and Vancouver have joined the municipal open data movement. They show vision in recognizing that providing detailed information about their cities is good for their citizens, the environment and the local economy. They should be feted for their leadership. Thank you, Canadian cities, for believing in Open Data.

Taking the plunge into open data has got to seem like pretty cold water for municipal governments. They offer up their data, in some ways an intimate look at the soft underbelly of their cities, for all of us to see, to analyze, to combine with other data. Nobody is paying to access this data. There is no direct cost recovery. All that the city gets in return is complaints: How few day care spots? How long to repave that road? Now I can prove that we need more bike lanes!

These complaints really are enthusiasm about and encouragement for our cities. Don't let them get you down. Still a left turn lane at that intersection would be nice, do you think you could... ?

Even with the community feedback regarding the data, it is possible and necessary to get the data into the hands of even more users from even further afield. The current license, somewhat shared by these three cities, only works at a superficial level and on a city scale. We can do better once we recognize the weaknesses in the license. This can be fixed.

EdmonTorCouver data license analysis

• No version number

  Without a version number there is no simple way to tell if the terms of the license have changed. Those interested in using data sets or updates must read the license again with every visit to the web site. Even the most successful, most widely adopted, most carefully drafted licenses in the Open space change over time. They have to change as the world around us changes. As examples,

  • Much of the software that runs the internet, companies like Google, organizations like Wikipedia and institutions like the London Stock Exchange, run on software licensed under the GNU GPL. GNU GPL has had three versions since 1989
  • The Creative Commons has licensed more than 350 million creative works since 2001. In the same time the Creative Commons Attribution license has reached version 3.
• Future Changes to Datasets/Terms of Use: The City may [...] change [...] Terms of Use. [...] Any change is effective immediately [...]

  This needs a better notification system for those interested in complying with the license and changes to the license. Please provide at least a data feed and a mailing list for notifications of license changes. How do future changes affect data acquired under an earlier license? Also see without any further restrictions below.

• License Mutation

  The licenses shown for Toronto and Vancouver differ in parts. (See No Warranty with Datasets and Exclusion of Liability and perhaps others) This evolutionary change, or mutation in the license may be beneficial, harmful or neutral, but it makes the licenses different. The Edmonton license was off-line with a 404 error when this was written. Do you think that minor differences like these don't matter? See the 18-month long Rogers case that involves a single comma and over $2 million.

• Authority: You agree to these Terms of Use by either: (1) clicking to accept the Terms of Use; or (2) downloading or using any of the datasets.

  Not required for a grant of rights. The user has no rights, beyond fair dealing, to data belonging others. Not a show stopper, but critical when combined with distribution, below.

• If you distribute [...] these datasets to any other person [...] you agree to [...] ensure any such person [...] is bound by, them

  This is a show stopper in several ways. Let's look at the this point when combined with the earlier points.

  • License mutations and any such person [...] is bound by, them
    The license when the data was acquired may have changed before the data is presented to a new user. Which license must be presented to the new user, the license when the data was acquired, or the current license at the city url?
  • No license version and any such person [...] is bound by, them
    Complying with this, if possible, is made much more difficult without version numbers for the licenses in question.
  • clicking to accept the Terms of Use; and any such person [...] is bound by, them
    License compliance requires downloading the dataset, or clicking to accept the license as listed in Authority. Presenting a visualization to a user does not require downloading the dataset. So click to acknowledge is required. Is a click required for each dataset used in the visualization, or for each license used in the datasets? Even a single click to acknowledge is a tremendous disservice to users, a usability issue for unexpected and creative use of the data, and a blocker to use of this data license in some, perhaps many, projects.
  • License mutation and any such person [...] is bound by, them
    At a minimum the Toronto and Vancouver licenses are different (see License Mutations, above). Presenting a map of Canada with this municipal data to a new user requires the user to at least a click for each city. If multiple datasets per city were acquired under different license versions this problem grows exponentially.
  • Surely this isn't what you expected?

    

• [ ...] but without introducing any further restrictions of any kind.
  This term is fatal to wide adoption of data under this license is several ways. You may find some of the prohibitions unexpected and unintended.
  • This data may not be combined with data under any other license, except public domain. Other licenses will have other restrictions.
  • This data may not be combined with data from other cities. Click to accept the Toronto license is a restriction, click to accept the Vancouver license is a restriction. Even if the Edmonton, Toronto and Vancouver licenses were exactly identical and of the same potential version number, they may not be used together.
  • This data may not be combined with data using a previous version of the license with additional restrictions.
  • This data may not be used behind your company firewall. A restriction to access is a further restriction.
  • This data may not be used by an assaulted person's help line if they want to password protect the location of their safe houses.
  • This data may not be used with data that is restricted to non-commercial use.
  • This data may not be used on a web site that is restricted to users of Microsoft Windows Internet Explorer.
  • This data may not be used in projects with a share-alike license.
• Giving City Credit: you may be required to remove a credit

  Credit being optional makes sense. But potentially demanding that credit be removed for the public interest just seems silly. A cautious developer will simply omit credit. This is not what you intend.

• Liability for Not Complying with Terms of Use:
  This section shows a lack of faith in the disclaimer of warranty and exclusion of liability. This section is a show stopper and blocks acceptance of the license by any sentient being. No this isn't about not wanting to comply with the license. Any sensible risk evaluation precludes accepting a clause like this. Quoted here in full for your entertainment:

  If, as a result of your breach of these Terms of Use, the City gets sued or is required to pay someone money, you agree to protect the City and reimburse the City for everything which you cause the City to suffer. This means that you agree to defend, indemnify and hold harmless the City and all of its officers, employees, representatives and agents from any and all liabilities incurred in connection with any claim arising from any breach by you of these Terms of Use, including reasonable legal fees and costs. You agree to cooperate fully in the defence of any such claim. The City reserves the right to assume, at its own expense, the exclusive defence and control of any matter otherwise subject to indemnification by you. You agree not to settle any matter without the written consent of the City.

• Governing Law and Jurisdiction
  This is another blocker to combining datasets from different municipalities. This also fails to recognize that global data sets and distributed projects may have servers in many different jurisdictions.
• Cancellation for Non-Compliance: The City may [...] cancel or suspend your access to the datasets without notice and for any reason, including anything which [...] is otherwise unlawful or harmful to others.
  Cancellation for non-compliance with the terms of use seems reasonable. The is otherwise unlawful or harmful to others is a problem. Unlawful in which jurisdiction? Harmful in what way? If I print a portion of the dataset and roll it into a tube, then light it and inhale, is that a smoking bylaw infraction? Or does that mean you'll cancel for non-compliance? If something is outside the terms and conditions it is outside the terms and conditions.

Another plea for the adoption of PDDL

The right license for municipal data sets is the Public Domain Dedication and License (PDDL).

The PDDL is published by the Open Data Commons at the Open Knowledge Foundation by lawyers expert in international law and database law. They licenses are drafted in consultation with Open Data communities around the world including the OpenStreetMap Foundation.

I'll write about this in more detail in a few days, but the advantages of the PDDL are many, including these:
• The license is drafted and maintained by subject matter experts.
• The license is drafted and maintained at no cost to the city.
• The license is drafted and maintained in consultation with the largest Open Data communities in the world.
• The Open Data Commons provide version numbers and notification channels.
• PDDL data is available for combination with the widest possible variety of other data sets and licenses.
• The PDDL disclaims liability for the city and publishes the data as is.

Conclusion

The data license used in Edmontorcouver is fatally flawed and prevents use in many of the ways that use was surely intended. Law is hard. Database law is hard. International law is hard. But Open Data is the right thing to do. Municipalities don't belong in the Open Data License Writing business. Municipalities belong in the Publishing Open Data In The Best Interest Of Their Citizens business. Let the Open Data Community help you do this.

Related articles

http://weait.com/content/tragedy-edmontorcouver-open-data
http://weait.com/content/better-approach-municipal-open-data
http://weait.com/content/another-approach-open-municipal-data

Credits

Lighthouse photo CCBY Dave Bleasdale on Flickr.

Map image and data © 2010 CCBYSA OpenStreetMap and contributors. The map image on this page includes no data from the Edmontorcouver data sets as the license was unacceptable at the time of writing.