Dear all,
TL;DR: Today the Wikimedia Foundation published a press release[1] about the “Wikimedia Enterprise” API project - announcing our first set of customers, as well as a new self-signup system. This is a significant milestone because it fulfills several promises we have made to ourselves and to the movement. There will be a community open meeting on Thursday 23 June @ 1700 UTC as documented on our Meta page.[2] The text of this email is also published on the talkpage - please centralise any comments/feedback there.[3]
Details
I am writing today with details of the latest developments in the “Wikimedia Enterprise” API project. This follows the project’s community-discussion phase, which began approximately one year ago, focused on Meta [and also this wikimedia-l thread]. Then, this past October, we issued a press release announcing that we were “open for business” on the project’s new site.[4] This wikimedia-l email thread contains the details of those previous phases.
Now is the third and final major announcement in this journey from “idea” to “reality”. Today’s press release,[1] and associated story on the project’s new “news” page,[5] states that:
Two well known organisations will be announced as the first customers of the project. One is a major social/search corporation [Google], as our first official paying customer. This also means that the project is now covering its current operating costs. The other is a movement partner and nonprofit organization [The Internet Archive] that will receive access at no cost.
Anyone will be able to sign up for an account and use/access the service [but not at a commercial scale] for free. Furthermore, payments for usage above that threshold will be calculated simply and publicly based on the number of API requests and gigabytes of data used. (Other free access methods for the dataset continue to exist, as documented on our Meta FAQ)[6]
The API’s metadata has been expanded to include the beta version of what we are calling “credibility signals”. This is already public information (such as pageviews, edit-rates, and page-protection status changes) packaged within the single data feed to help users make more informed decisions about when they should refresh their copy of the dataset. (Emphasis on ‘beta’, as this is not available on all versions of the product yet.)
This announcement is a significant milestone because it fulfills several promises we have made to ourselves and to the movement, namely that:
We have built something that commercial organisations who are already heavy users of Wikimedia content and Wikimedia Foundation services are willing to invest in. The pricing is based on estimated usage, resulting in a more manageable and transparent cost structure. The project is now covering its current operating expenses. In addition, we requested and received a public affirmation/support letter from the Board for the project’s financial operating principles, ensuring that commercial revenue will only ever be a minority of the total and their oversight for any future high value contracts.[7]
The nonprofit partner will receive access at no cost, demonstrating a first practical example of how this project supports the mission of knowledge access while also providing a new revenue stream.
The ‘trial’ tier of the service is primarily designed to allow potential customers to determine whether they want to use it in commercial production environments, but it also allows anyone to see what is ‘in’ the API. Moreover, it will allow volunteers or researchers to access the service for free at a non-commercial scale. If those people have a mission-relevant use-case that requires them to continue to use the Enterprise API above that scale (i.e. that isn’t viable using other APIs/dumps), we will continue to provide them with free access.
The ‘credibility signals’ concept means that vandalism and errors should appear less often and/or be removed more quickly in downstream services such as search engines. [Note, how and when customers incorporate features is their own decision].
While we are proud to announce these customers, it is important to note that our market research has identified a significant gap in our movement’s ability to have Wikimedia knowledge used. The world’s largest companies are already using Wikimedia; we’re just providing a better way for them to do so. But for everyone else, it is often too hard and they do not have the resources (financial, technical, and human) to incorporate Wikimedia information – even though they want to. In short: simply providing legally-reusable knowledge is insufficient to enable reuse for a very large portion of society.
And so, we are focusing a lot of our future product development on this Knowledge as a Service model - consistent with the Movement Strategy’s “strategic direction”.[8] This is what OpenFuture.eu’s interview with the Enterprise team referred to as our attempt at “lowering the playing field” – a term we quite like.[9]
We are increasingly realising that the future of Wikimedia Enterprise is much more nuanced than merely “making big tech pay”, but is about enabling access to the many companies who want to use Wikimedia knowledge in their own products but currently can’t. These organisations are willing to pay us to find ways to better support their specific use of Wikimedia content, both through more accessible technology, contractual guarantees of service availability, and professional services to help them make the best use of our content in their systems. As per our enumerated Principles,[10] all customers get the same product - there are no exclusive or bespoke features - they only pay for the volume of usage. This will allow smaller companies to compete and will ensure that Wikimedia knowledge is more widely available. Our goal is for the future business model of Enterprise to resemble “many paying a little” rather than “few paying a lot” – an approach similar to our movement’s “many small donors” fundraising methodology.
Still to come later this year will be:
Exploring options to integrate Wikidata in the dataset, which is a common customer request. We are working closely with Wikimedia-Deutschland to discuss how to best do this.
Small, and non-U.S. based customers. This is crucial to demonstrating the Knowledge as a Service value of the project. We already work with relevant Chapters when we have a potential ‘local’ customer who has expressed strong interest.
Publishing aggregate revenue/expense data, but only after there’s enough aggregate financial data collected, over a sufficient period of time, and with enough customers to be informative.
The “news” page on the Enterprise website itself will be where future software updates, customer case-studies, etc. will be published. This ensures that the information is available, while not detracting from community-focused places like the Diff blog. The first post on that news page is available today.[5]
In order to not distract from community-focused discussions, and of course unless there’s something important for the community to know, in the future we will announce new customers product updates etc. on the project website’s news page and on Meta rather than on this mailing list etc. - but we felt it was important to do it this time.
Finally, I would also like to ask you to keep your eyes and ears open for anyone in your corner of the Wikimedia community who has questions or concerns about the project. Please ask them to read and comment on our Meta FAQ,[6] to contact me directly, or to attend the public community open meeting on Thursday, which is already announced on Meta on the project’s page & the mainpage calendar.
Sincerely, and on behalf of the Wikimedia Enterprise team,[11]
Liam Wyatt [Wittylama]
|
[2] https://meta.wikimedia.org/wiki/Wikimedia_Enterprise
[4] https://enterprise.wikimedia.com/
[5] https://enterprise.wikimedia.com/news/hello-world/
[9] https://openfuture.eu/blog/wikimedia-enterprise/
[10] https://meta.wikimedia.org/wiki/Wikimedia_Enterprise/Principles
[11] https://meta.wikimedia.org/wiki/Wikimedia_Enterprise#Team