Talk:Wikimedia Enterprise

Archives

---

• /Archives/2020
• /Archives/2021

Wikimedia Foundation Board Statement on Wikimedia Enterprise revenue principles

On behalf of the Board of Trustees of the Wikimedia Foundation.

The Board has followed the community conversation about Enterprise with interest. Thank you all for engaging in a productive conversation. As Board members, we reaffirm our support for the Wikimedia Enterprise project, and in particular the following principles relating to its future revenue.

First, the Board sees Wikimedia Enterprise as a supplemental revenue stream, not meant to replace fundraising as the primary way the Wikimedia Foundation is funded. To that end, revenue generated by the Wikimedia Foundation from commercial activities, like Enterprise, will not surpass 30% of the Foundation's total revenue in a fiscal year. The Foundation and Wikimedia Enterprise will not seek commercial revenue beyond that 30% limit. This is in line with advice from the Foundation’s auditors regarding the U.S. Internal Revenue Service (IRS) limits on commercial revenue earned by public charities. Additionally, revenue from Enterprise remains under the oversight of the Wikimedia Foundation. There are no plans to earmark profits from Enterprise for any specific program or for the Wikimedia Endowment.

Second, the Board sees the Gift Policy as a model for Board oversight and transparency for Wikimedia Enterprise revenue. The Wikimedia Enterprise team will notify the Board of all agreements expected to generate revenue in excess of $250,000 USD annually and allow the Board to raise concerns, should Board members have any. This is consistent with how the Wikimedia Foundation treats large corporate donations.

Finally, the Board perceives Enterprise as a step towards treating corporations, and being treated by them, more fairly. Large corporations, which rely on Wikimedia services heavily, and so far have not adequately contributed back, will have a clear way to do so; while smaller organisations, with fewer technical resources, will be able to get high quality service that they would not be able to obtain otherwise.

Thank you again for your engagement in the conversation about Enterprise. It is gratifying to see this new source of support for Wikimedia projects and movement taking form.

Best regards,

antanana / NTymkiv (WMF) (talk)

Acting Chair, Wikimedia Foundation Board of Trustees

P.S. Copied here for convenience from a subpage with the Statement --NTymkiv (WMF) (talk) 15:11, 11 October 2021 (UTC)

Update

[This is a version of a message also sent to Wikimedia-l]

A lot has happened since March when we Introduced the Wikimedia Enterprise API and began community conversations about the project’s development. Now is an important milestone to give everyone:

1. an update responding to community advice we’ve received; and
2. to describe what is happening next.

The idea of an API for the specific needs of the commercial sector had been discussed for more than a decade (both for the purposes of improving user experience, and also to diversify revenue). The announcement in March introducing the Wikimedia Enterprise API generated a lot of Wikimedia-community and mainstream-media attention - most notably in WIRED. Since then, the team has been hard at work building the actual product and hosting many conversations (regular public meetings and participation in events including SWAN, Wikimania, EMWCon, Clinic) - as well as a considerable volume of discussions on this talkpage (see Archives 2021). All of this has generated lots of suggestions, which we have endeavoured to incorporate and respond to before the actual commercial launch. On behalf of the whole team, I thank the many many people who have been willing and able to share constructive feedback with us over these months. Links to recordings from those meetings/presentations can be found on our meta homepage.

1. Updates in response to community advice

1.1 WMF Board statement.
Subsequent to the most recent WMF Board of Trustees meeting, a statement reaffirming their support of project, and in particular its operating principles relating to its future revenue, has been published. You can find the board Statement immediately above (transcluded from the Wikimedia Foundation Board noticeboard), and the Enterprise operating principles here. Consistent community feedback was that our published principles were good and sensible, but for such a new and unusual thing in our movement, an overt statement from the Board of Trustees was requested. This statement affirms that:

• revenues of the WMF obtained from commercial activities shall not surpass 30% of the total planned revenue via all sources (including donations) in that fiscal year, and no further revenue would be sought beyond that limit;
• the Board of Trustees will be notified in advance of any large commercial agreements - exactly mirroring the procedure for large gifts;
• revenue obtained from Wikimedia Enterprise services is under the oversight and control of procedures for revenue raised by the Wikimedia Foundation and the revenue will not be earmarked for a specific program.

Each of these things were already noted on-wiki, but they were very much worthwhile re-stating formally.
Equally it bears repeating: That the existing APIs and methods of accessing Wikimedia sites remains. The creation of this optional commercial service, designed for those with specific high data-volume demands, does not change the experience (legally or technically) for anyone else.
Relatedy, and also in response to community suggestion, the formal contracts which define the legal relationship of the non-profit Wikimedia Foundation to this commercial activity, have now been published on the governance wiki and linked from the relevant section on the FAQ page.

1.2 Free technical access for the community.
You will soon be able to access a copy of the Enterprise dataset, refreshed each fortnight, at the Wikimedia dumps portal. Furthermore a ‘daily dump + hourly diff’ version is also already available, via Wikimedia Cloud Services to any users of Toolforge, Cloud VPS, or PAWS. Both of these are provided to anyone, for free (in both ‘gratis’ and ‘libre’ senses of the word). Importantly, and consistent with community feedback, neither of these access methods require any special request process to access them, other than the existing terms of service on those platforms. Software development updates are published monthly on our page on MediaWiki (as is the API documentation), and the work is coordinated on our Phabricator board.

1.3 Next public meeting.
As mentioned, the team has been holding regular public calls. If you would like to meet with, and ask any questions of, the Enterprise team (a.k.a. “Office hours”): Friday October 22 @ 1500 UTC on Zoom. If you would like to arrange a conversation about this project with a group in the community that you are part of (at a time, language, and meeting-software platform of your choice), please contact me directly.

2. What is happening next

The launch of the project’s standalone website, denoting the service as “open for business”, will take place early next week on Monday 25.

In parallel we will also be announcing the first customers Separately, we will announce the first set of customers at few months' time - both from the commercial and non-profit sectors. We expect this will generate some attention in the media. Despite our best efforts to be visible across the wikiverse, reading about Wikimedia Enterprise API in the media will probably be the first time that some Wikimedians hear about it - which might be surprising for them. So, next week, if you see any Wikimedians asking about this project on community forums, please notify me by email, on-wiki, and/or direct them to this page or the the FAQ (currently available in 7 languages).

Once again, thank you to all who have been involved in the development of, or given feedback to, this project.

Sincerely, LWyatt (WMF) (talk) 16:57, 11 October 2021 (UTC) + Edits made on 08:47, 27 October 2021 (UTC)

The press release published on the 25th is here, and the project website (as already announced from the Infobox on this Meta page) is enterprise.wikimedia.com. LWyatt (WMF) (talk) 08:47, 27 October 2021 (UTC)

What exactly does payment obtain?

I'm looking through the service descriptions, and the best I can tell is that the only unique offerings for paid-fee services are "real time" access, and of those only the "firehose" of all projects' recent changes is substantially different than what can be obtained from Syndication feeds for example. Is that correct?

I suspect that, except for the very largest customers, there will not be much use in practice for the "firehose" beyond keyword spotting, resulting in rapid reversion of anything showing the user in a negative light. If that turns out to be a substantial use, is it not equivalent to enabling a new form of paid advocacy editing? Ta76549876 (talk) 18:54, 11 October 2021 (UTC)

Other, non-software, "new" aspects of this service are the fact of the SLA (service level agreement), and customer support. They sound simple but should not be underestimated as needs from a commercial sector perspective. The SLA is the legal guarantee - the contract - that the service will be available and working in the agreed way with the very high 'uptime %' that is required by these kinds of organisations. Having a '99.99% guarantee' is something that companies rely upon in other major API services, and is something we've never been able to offer - because that kind of legal promise is not something that our infrastructure is designed to formally support. Equally, that means if you want to change anything, you need to tell them in advance. Having a separate, dedicated feed, that has explicit infrastructure design requirements to serve that need - is far too 'heavy' for what normal wikimedia uses require.

Secondly - the customer support aspect is fairly self-explanatory. Finally, the API is being developed based on the software-roadmap feature requests from these companies (taking into account our Wikimedia Enterprise/Principles that it does not include any "exclusive" data - it's not different data, just formatted the way they want it). This means that the existing APIs can focus on the needs of 'normal' users and not have to think about whether it would affect the very high-volume users, and, equally, we can develop this service focusing on what the commercial organisations want, without fear of affecting something that only volunteers use. Win-Win.

Finally: This is a "read" API - not a "write" API. You can't edit Wikipedia with it. See also the FAQ section "Will it directly affect Wikimedia content?" - the answer is: No.

I hope this answers your concerns Ta76549876. Sincerely, LWyatt (WMF) (talk) 19:17, 11 October 2021 (UTC)

I'm not sure what this means: Having a '99.99% guarantee' is something that companies rely upon in other major API services, and is something we've never been able to offer - because that kind of legal promise is not something that our infrastructure is designed to formally support. What does it mean for the infrastructure to formally support a legal promise? What happens if the WMF doesn't meet its SLAs? And lastly: why is it not a goal to achieve this same high standard of reliability for our (non-corporate) readers? --ATDT (talk) 04:02, 13 October 2021 (UTC)

To question 1 - if we don't meet the SLA requirement we have a customer, their contract with us will determine what happens. Generally speaking this is broken into % ranges - e.g. missing an uptime obligation by thousandths, or hundredths, or tenths, or whole percentages - and the primary remedy is refunds/future-discounts. This is standard in the industry.

To question 2 - The formal requirement of this kind of uptime for Enterprise means we have to have a lot of redundancy and capacity in the system to be absolutely certain we can meet that responsibility. To build that kind of "slack" in the system for the existing infrastructure would be wasteful of resources and also very time consuming to set up. Wikimedia's existing services are already extremely stable and reliable - especially considering how small our infrastructure is compared to other organisations that run websites at our scale - and they do aim to achieve high reliability for everyone, and they achieve it! The difference is that the WMF is not legally liable for making large refunds to "customers" if Wikipedia stops working for 5minutes. That would be a huge financial risk for the existing service to burden itself with. This is also extremely related to why Enterprise is - during this set-up phase at least - being built on an external cloud provider, not in-house infrastructure. You can read about that in technical detail at Application Hosting on our MediaWiki page. When we are up, have a stable customer group, and comfortable with what kind of usage volumes are required... then we can look at potentially moving that hosting in-house. But to do that from the start would require extremely expensive and slow process of hardware commissioning for a currently-unknown service requirement.

I hope that covers your questions, ATDT. Sincerely, LWyatt (WMF) (talk) 14:22, 13 October 2021 (UTC)

Staff

I read in Mediawiki that the staff who works on the Wikimedia Enterprise project does not directly works for the Wikimedia Foundation. There is an agency used for the tasks in that case. How much is the additional amount of money that is needed, because the people work not directly for the Wikimedia Foundation. After my understanding usually a service is more expensive as doing it on the own. Also I understand there are difficulties if you have the staff directly at the Wikimedia Foundation and you cant be sure how long you need the staff. Please think about if you can employ the employees directly at the Wikimedia Foundation. I think that is better and propably cheaper.--Hogü-456 (talk) 20:08, 15 October 2021 (UTC)

Dear Hogü-456, I assume you saw me add a series of "credits" to the bottom of the Wikimedia Enterprise#Team section last week - noting the various non-WMF organisations who are contracted to provide specific services to this project. Although it is not common (and not a legal requirement equivalent to copyright attribution) I thought it was appropriate, fair, and kind, to specifically acknowledge those teams' work. In adding this series of "credits" I also removed the sentence which was there before, which partially answers your question:

"At this early stage in the project, we are not yet sure of the long-term engineering needs and we wish to thoroughly assess the project's ability to become self-sustaining. This way we don't excessively disrupt other WMF projects or divert resources."(diff).

That is the human equivalent of why - as documented onour mediawiki page in the "application hosting" subheading - we are building the technology on an independent system: So as to not place a burden of work upon the WMF's existing staff or infrastructure for this new, and very different, purpose. When, in the future, we are more confident what the size/scale/type/frequency of work is required, we can carefully transpose that work (and infrastructure) in-house.

Another aspect is that, hiring permanent staff at the WMF is a slow process because it is important that it is done carefully and appropriately. Of the people named on that "Team" list - only myself and User:NNzali (WMF) were already WMF employees before this project. The fact that the team now consists of 8 staff within slightly over a year (and we're still recruiting) is extremely fast in this context. Nonetheless, there are constraints that mean we do require external contract work for several overlapping reasons: Some things are required, but only required to be done at the beginning/once (e.g. designing the new website [which will launch next week]); Some things which are required are not within the standard expertise of existing Wikimedia Foundation teams (e.g. "sales"); And some things needed to be done much faster, and required more people to do them, than could be reasonable hired for in a short space of time (e.g. the development of an commercial-scale API set that can withstand the technical needs of the world's largest companies - within a year)!

So yes, I completely agree that where possible, work should be done by Wikimedia Foundation staff, but there are legitimate reasons to do so in several circumstances - especially at the "startup" stage - that are also the most prudent financially. LWyatt (WMF) (talk) 16:17, 18 October 2021 (UTC)

Pricing

Hello,

can you please tell me how the mentioned prices at the website are calculated. I can find there two numbers 25000 und 50000 Dollars. I thought that there is a more detailed pricing system and so big consumers pay more, if they use the API more. Also I think it is good if you record or analyze less data and so maybe a simple pricing model is better.--Hogü-456 (talk) 19:28, 27 October 2021 (UTC)

Dear Hogü-456 - the two prices mentioned on that page are the "start" prices for those different "products". Respectively: "Bulk Data Feed Delivery" corresponds to the equivalent of the Dumps that we are also providing for free every two weeks here: https://dumps.wikimedia.org/other/enterprise_html/. And "On-Demand Access" corresponds to the Daily Dumps+Hourly diffs version that we are also offering for free via Wikitech:Portal:Data Services.

With the commercial service, as implied on that pricing page, there are many other factors involved in what creates the price for any individual customer. These include and not limited to: Which of these two services are wanted (or both, and/or also the "real-time" feed service); the maximum requests per minute/week/month; up-time % requirement per month; customer service access 24/7 or less; the number of days notification you require before we make any updates/new-feature to the service; and then any potential discounts for choosing to combine any of these things (or perhaps because of mission-alignment or non-profit status); the duration of the contract itself; etc.

As you can see - there's lots of "moving pieces" and much to negotiate! And therefore, it's impossible at this early stage in the project to publicise a clear price list for all these potential options. Most certainly, the biggest companies - those which require the most stringent/extensive versions of the criteria I listed above - are necessarily charged a higher amount.

Over time, as we negotiate contracts with more customers - large and small - we hope we will obtain a much clearer understanding what are certain "standard packages" of all these things that different groups of organisations want. When that happens we will - as you recommend - be able to make (and publish) a simpler pricing model that future customers can select from. A kind of "menu". LWyatt (WMF) (talk) 21:50, 27 October 2021 (UTC)

Packaged metadata exclusive to Wikimedia Enterprise

The website mentions "packaged metadata exclusive to Wikimedia Enterprise". What kind of metadata are we talking about? --Andreas JN466 22:14, 27 October 2021 (UTC)

The metadata which is all the existing things in Wikimedia sites - like whether the page is “protected” or when it was created/deleted/last edited - but brought together in a consistent format across all languages/projects. The information isn’t ‘new’, the grouping of it is. LWyatt (WMF) (talk) 22:19, 27 October 2021 (UTC)

Does this include the page history? ---<(kmk)>- (talk) 04:20, 28 October 2021 (UTC)

That's strange, as metadata mostly refers not just to technical aspects of a piece of information, but rather to what it means, see subject cataloguing in libraries. Regards, Aschmidt (talk) 05:29, 28 October 2021 (UTC)

Dear Aschmidt, The "marketing" text on the website is necessarily brief/simple - so "packaged metadata" was the easiest way we could describe it. Another term we are using when talking with potential customers is the idea of "credibility signals". If you look at the Enterprise MediaWiki page, which has the technical documentation, and do a text-search for "credibility signals" you can see where this term is used/described. In summary: re-users of Wikimedia and especially of Wikipedia (e.g. Search engines) want to be able to have better understanding of "is this vandalism?" [broadly defined]. Obviously, inside Wikipedia(s) we have lots of different ways of notifying/identifying this to ourselves but there is no single yes/no sign. Instead we all look at things like "page protection status", or "are there pending revisions", or "the size of the bytes changed in the last revision", or "is this article suddenly getting a lot more edits than normal"etc. The Enterprise data feed is trying to package all these things in a consistent way to make it easy for a customer to chose which "signals" are important to them so they can decide how to use the data. We are not creating/inventing any "new" signals that are exclusive to paying people - all this information already exists - but we are "packaging" it in a way that the high-volume users can more easily ingest into their own products. To emphasise, we are not creating "a vandalism free version"; instead, we are [hopefully] making it easier for large reusers to make their own decisions.

If you are interested - you can always download a database dump and play with the information yourself - here https://dumps.wikimedia.org/other/enterprise_html/ . The product roadmap is here: Wikimedia_Enterprise#Product_Roadmap

To User:KaiMartin - There are three different services: an "on demand" version (which is equivalent to a database dump of the whole current version); a "bulk data feed" (which is equivalent to a database dump + a "diff" of all changes since the last period e.g. hourly); and a "real-time data feed" which is equivalent to EventStream (where any change is notified immediately). LWyatt (WMF) (talk) 08:41, 28 October 2021 (UTC)

Dear @LWyatt (WMF), thanks for elaborating. I just would like to add that what you've described may be some kind of metadata, but when we speak about metadata we usually mean descriptions that tell you what a piece of information actually signifies. This morning I mentioned subject cataloguers in libraries who describe what a book is all about with subject headings and a DDC notation, etc. Creating metadata is hard work performed by experts, it's also quite expensive. There is machine-created metadata, but it's rather bad and I understand the WMF does not play on this field dominated by linguistics and AI. I am sorry to say that what we find on the Enterprise website in this respect is misleading. Regards, Aschmidt (talk) 19:54, 28 October 2021 (UTC)

Aschmidt, as someone who comes from the museum and library field, I certainly understand that formal metadata creation - of the kind like my former colleagues who are cataloguers at the National Library of Australia do - is a complex, precise, timed-consuming, and manual process. The structuring of high-volume and 'messy' data Wikimedia generates into a feed which is more machine-readable is not the same kind of activity at all. We certainly agree! Nonetheless, the data about the contents of Wikimedia pages - e.g. it's edit-protection status - is a kind of metadata and I don't think there's an alternative word for that. Given the constraint that we can't go into high-detail on the enterprise website itself, can you think of a better and accurate way to introduce the concept we are discussing? Can you propose a better phrase than "packaged metadata"? I would like to be able to describe "credibility signals" but that is definitely too complex for a promotional website - and is more appropriate for the technical documentation. Do you have an alternative suggestion? LWyatt (WMF) (talk) 08:19, 29 October 2021 (UTC)

Thanks for your reply, @LWyatt (WMF). I am glad to read that we both have a library background, so we both know about the meaning of the term metadata in a more elaborate environment than this. The website even calls the metadata Wikimedia Enterprise offers packaged metadata. This is an altogether unusual term, as there are plenty of ways to package something for a particular purpose. This says neither what has been packaged nor what the package should be fit for. I think we could put it straight by naming it technical metadata that are somehow fit for further processing. So it becomes clear that you do not offer subject metadata, but data about technical aspects of Wikipedia pages or about the edits in Wikipedia. You might also like to give examples for what you can do with these data. What do you think about that? Regards, Aschmidt (talk) 16:04, 29 October 2021 (UTC)

@ Aschmidt, I certainly agree that "packaged metadata" is unusual - and in this usage it is probably a neologism. I'm not convinced that "technical metadata" is better - because this would also suffer the possible critique from other people that "all metadata is technical" and that this phrase is also unusual. Nonetheless - I personally do agree with you that "technical" is better than "packaged" and I will pass your suggestion to my colleagues. As for including examples: yes, that is the intention. We plan to make some kind of "case studies" page on the website which describes how some different customers (large and small) are using the service in different ways for their own purposes. This would certainly feature descriptions of what this Metadata is, and how it can be utilised in practice. As we have only just launched we do not have these case studies available yet, but it will come. LWyatt (WMF) (talk) 11:33, 31 October 2021 (UTC)

@ Aschmidt: To follow-up on our discussion. I forwarded this issue to our designer, and they agree. The new text on the main page of the website now says "...access metadata packaged exclusively for Wikimedia Enterprise...". This has two advantages: it removes the qualifier word before "metadata" (since neither "technical" nor "packaged" were adding value to describe the word "metadata" better, AND we've moved the word "exclusive" to a different part of the sentence (this emphasises that the metadata isn't exclusive/new, rather, it is the package itself). I hope you agree with me this is an improvement. As mentioned before, in the future we intend to add practical "case study" information which gives explanations of how/what the metadata can do for potential users. Sincerely, LWyatt (WMF) (talk) 19:05, 2 November 2021 (UTC)

Is the version history submitted and how is it made sure that the license conditions are kept, at least if the data is not CC0? --Ailura (talk) 09:43, 29 October 2021 (UTC)

@Ailura, I'm not quite sure I understand the question in the legal or technical detail you are requesting. But, you can always investigate the dataset yourself, downloadable here: https://dumps.wikimedia.org/other/enterprise_html The license information [CC By-SA, since neither Wikidata nor Commons content is included in the service - yet] is included - in a manner that is legally-equivalent to the way that is already happening in, for example, the EventStream. Nothing is different in that regard. LWyatt (WMF) (talk) 09:50, 29 October 2021 (UTC)

@LWyatt (WMF), I think @Ailura means to ask whether the metadata you provide include an attribution and licence information as can be seen with each edit in the version history of a Wikipedia page. This information should not be lost in the technical process. Regards, Aschmidt (talk) 16:11, 29 October 2021 (UTC)

Thank you for this clarification Aschmidt. Yes Ailura, the attribution information IS included. You can see the explanation of all the different fields provided in the Metadata at the moment at mw:Wikimedia_Enterprise/Documentation#Data_Dictionary. This may grow in future versions. The editor's name is included in the version.editor field. LWyatt (WMF) (talk) 11:36, 31 October 2021 (UTC)

Are there discounts for companies that don't track users?

Will companies which, like Wikimedia, commit to not tracking their users (DuckDuckGo, say) pay less for the service than a Google, Facebook or Amazon? Or is the pricing entirely independent of that? --Andreas JN466 22:15, 27 October 2021 (UTC)

Mission-alignment can indeed be a factor. LWyatt (WMF) (talk) 22:19, 27 October 2021 (UTC)

Thanks, good to hear! --Andreas JN466 22:54, 27 October 2021 (UTC)

@User:LWyatt (WMF): Can be a factor or will be a factor? Is the pricing going to be public? ---<(kmk)>- (talk) 04:23, 28 October 2021 (UTC)

KaiMartin - IS a factor, among many others. I've made a more detailed answer about pricing factors on this page, see #Pricing above, and the short answer is that there is no fixed price 'menu', at least not yet. But yes - "mission alignment" is something that is an important consideration - because this new service is designed to respond to two points in the Movement Strategy. As described in the FAQ document, this service is about "increasing the [financial] sustainability" but ALSO about "improving the user experience [for down-stream readers]". That means that Enterprise supporting the services which are aligned with Wikimedia's mission is a direct strategic goal. "Supporting" in this sentence could refer to discounted pricing, but could also refer to other kinds of collaboration. For example: the original question mentions DDG - last month the results of a professional research collaboration with them was published on the Wikimedia Diff blog: "Searching for Wikipedia: DuckDuckGo and the Wikimedia Foundation share new research on how people use search engines to get to Wikipedia". That was not an Enterprise project, but is a recent example of non-financial collaboration and "support" with the for-example organisation in the question.

To your second question, the individual contracts with each customer will not be published but the aggregate data will be, and all major contracts will be reviewed in advance by the WMF Board. You can see details about financial transparency/oversight policies on the FAQ.

I hope this answers your questions. LWyatt (WMF) (talk) 08:18, 28 October 2021 (UTC)

Will an Indian or South African company have to pay the same as a US one?

Do prices for these services differ in different regions across the world? For example, will an Indian or South African company entering this market have to pay the same as a US one? --Andreas JN466 22:53, 27 October 2021 (UTC)

See my earlier comments on the preceding subheadings you created. The short answer is "it would depend if mission-alignment is a factor" and "it depends on the kind of service volume and frequency (and customer support) they want". There is no automatic calculation based on the GDP of a country where the company is registered, if that's what you're asking. LWyatt (WMF) (talk) 08:20, 28 October 2021 (UTC)

I was, kind of. Don't you think some thought should be given to this? The WMF is talking about "knowledge equity", and about growing WMF revenue to perhaps as much as a billion dollars a year so we can "help the global south" draw level.

If we are really serious about helping the global south help itself, then charging people there less would make a lot of sense to me. Otherwise you have a situation where either –

1. only first-world companies will be able to afford the service (ensuring their dominance of Global South markets and the continued flow of wealth from these countries to the US, compounding the existing problem of inequality), or
2. you will have to offer grants to the Global South so they can afford the expensive service ... that combination of extracting wealth from less-well-off countries with one hand while giving "philanthropic aid" with the other hand is a well-known dynamic that's rightly been criticised and that we should not replicate.

Thoughts? --Andreas JN466 12:55, 28 October 2021 (UTC)

As I said earlier, but will restate: it depends. Negotiations with each potential customer must be taken on a case by case basis, at least at this very early stage... Just because a potential customer has its business registered in a 'global south' country does not automatically imply it has any greater mission-alignment to the Wikimedia movement. And vice-versa, if a potential customer is registered in a 'global north' country does not automatically mean they are less deserving of special consideration. LWyatt (WMF) (talk) 13:13, 28 October 2021 (UTC)

Will there be a publicly available list of clients, so people can see to what extent the companies whose products benefit from this service are distributed across the globe, or are regionally concentrated? --Andreas JN466 13:36, 28 October 2021 (UTC)

We hope, and and indeed we expect, that new customers will not only be able to be listed in some manner, but that they will actively want to announce their being a customer. Eventually we hope to build a "case study" section on the website to show future-customers the diverse kinds/sizes of organisations that can benefit from this service, their diverse industries and - as to your specific interest in this question - their different 'home country'. Of course, this does not yet exist yet, but will come in time.

Depending on what makes sense at the time we will not necessarily make a new announcement, or an immediate announcement, for each individual new customer. For example, some customers may make their own announcement on their own site (that we re-share), or vice-versa; or, some may wish to not make any loud news but we just mention it somewhere. Or, we may make some kind of "group" announcement of several at once - that is likely to be the approach we take for announcing the first customers. Furthermore, I should add that Wikimedia Enterprise does not (yet) have its own independent social media presence, and I think people would get bored/confused quickly if the existing WMF press/social media platforms were being used each time there was some Enterprise news. Thus: the how, where, when new Enterprise news is shared is not yet defined. Perhaps it will be appropriate for us to make a "news" tab on the website (combining information about software version releases, new customers, appearances in the press or at conferences, etc.).

Finally - I should add that, especially for small customers, there is no legal obligation that a customer of Enterprise has to publicly say that they are a customer. I fully expect that all future customers will want to publicly say that they are a customer, (especially since they can already "secretly" get the same content via the existing Wikimedia APIs, which don't even require a signed API Key). Nonethless, it is legally the case that small businesses - like donors to the WMF - remain anonymous if they prefer. But - as noted in the Board Statement, all major customer contracts will be reviewed in advanced by the WMF board - exactly equivalent to the processes already in place for large donations. LWyatt (WMF) (talk) 14:25, 28 October 2021 (UTC)

Thanks for your comprehensive answer, Liam. --Andreas JN466 15:37, 28 October 2021 (UTC)

Rataoksr

The excellent squirrel on the website's front page is apparently Rataoksr, a very friendly squirrel who has a habit of carrying slanderous gossip between willing parties - this isn't really a question, but it seemed very apropos for Wikipedia ;) Nosebagbear (talk) 11:33, 28 October 2021 (UTC)

That was the original pseudo-etymology that we gave to the Squirrel-logo - based on the Norse mythology - yes. But, after discussion with some Scandinavian Wikimedians, we decided to remove the name because we did not want to make a bad/simplistic appropriation of another culture's ancient stories. LWyatt (WMF) (talk) 13:18, 28 October 2021 (UTC)

That sounds fair, and a good call - and I should say that I really like the colour scheme and appearance, too Nosebagbear (talk) 16:36, 28 October 2021 (UTC)

wikimedia enterprise is normal business, what are the consequences?

a cool service that is. it is normal business like all other businesses repurposing contents, and will pay taxes. it has nothing to do with WMFs purpose stated in its bylaws, its vision and mission to provide free knowledge to everybody. for us as community this boils down to three main questions:

• is it mission that some of the wikimedia organizations own a stake in a commercial company or not?
• is it legal that 501(c)3 funds are used to build a paid service? in most european countries it would be illegal. like, the board of WMDE would go to prison if they transfer tax exempted money to WMF. the rule is black and white: "everybody has free access or it is not tax exempted".
• do we want that the WMF tries to monetize voluntary work or not? since its existence wikipedia always had the challenge that some people tried to monetize their contributions, or tried to monetize others work. wikipedia was unique to say "no, we do not monetize". if WMF now monetizes, it looses its magic factor, and becomes a normal NGO.

for the legal aspect - the most similar to this is a football club. only some people or companies have access to its services so it is not tax exempted. they are free to create as many companies as they want. i am pretty sure there is a grey area so even tax exempted organisations might own commercial companies. do you have some background info here? --ThurnerRupert (talk) 06:04, 29 October 2021 (UTC)

Dear ThurnerRupert. I will copy-past the answers here which I have already written to you where you asked me these questions first, on my personal facebook page:

To question 1: Yes. The project is directly correlated to two items in the Wikimedia Movement Strategy. Specifically “Explore new opportunities for both revenue generation and free knowledge dissemination through partnerships and earned income - for example...Building enterprise-level APIs” and making “the Wikimedia API suite more comprehensive, reliable, secure and fast, in partnership with large scale users where that aligns with our mission and principles” For more details see: FAQ#How does this relate to movement strategy?

To question 2: Yes. It is legal. It is extremely common for a charity to have some commercial activity and for that activity to be operated by a subsidiary. Some examples among organisations we are friends with include Linux, CreativeCommons, Mozilla, and Open Data Institute. For details of that context see Essay#How will this be structured legally? For details of the formal relationship between the WMF (a 501(c)3 organisation [American charity]) and this wholly-owned Limited Liability Company [LLC] - including the full text of the contracts themselves - see: FAQ#Why is this being run by a subsidiary?

Also, the The assessment of appropriate tax treatment of the LLC activities has been coordinated with the Wikimedia Foundation auditors KPMG - as per the minutes of the WMF Audit committee from July 2020.

To question 3: If we consider readers/users of Wikimedia content who are 'downstream' of our own websites (i.e. the millions of people who access Wikipedia via third-party services) to be worthy of our attention as 'readers', then it follows that we should be investing in their user experience. I think we should - and this is the point made by the second quote in my answer to question 1 - therefore the Strategy thinks so too. Their needs are valid. However, using the Wikimedia Foundation's existing financial resources to respond to these needs would mean subsidizing the software development and hardware costs of some of the world's largest commercial organizations with donor's money. I believe that an inappropriate use of movement resources. The Wikimedia Enterprise API avoids this through self-funding: where the commercial organisations who have specific, special, high-demand needs, can pay for the development and maintenance of those services - not donors. And furthermore, that those companies which make large profits based (in part) of volunteer labor can financially invest back into the Wikimedia movement - instead of the status-quo where they are extractive from it. For further details on this point, see: Wikimedia Enterprise/Essay#Self-funding

Finally, although you didn't ask specifically I will include that the financial oversight/transparency information is here: FAQ#How much money will this raise?. This project will never be permitted to raise more than 30% of combined annual revenue of the Wikimedia Foundation, meaning that the charity status, and also the fundraising style of the WMF will not be affected - our movement remains independent. And, the LLC is wholly owned - the staff (like me) are WMF staff and the money goes to the WMF, not of some spinoff company - so it sits with the same oversight and planning as all other things in the WMF.

Sincerely, LWyatt (WMF) (talk) 08:25, 29 October 2021 (UTC)

thanks a lot liam for the clarifications, very helpful! --ThurnerRupert (talk) 11:27, 1 November 2021 (UTC)

How will compensation of staff and directors be handled?

You say in the essay that LLC staff will always be WMF staff. I understood this to mean that you are committing to an arrangement whereby –

• the LLC itself will never pay any individual (including any LLC directors) a salary or any other type of personal compensation,
• will only ever pay fees for services provided by the WMF or outside firms who then compensate their staff,
• and all WMF people will always be compensated by the WMF and be subject to customary Form 990 compensation reporting.

Is this correct? --Andreas JN466 15:35, 29 October 2021 (UTC)

1) While it is legal for the LLC to hire its own staff... That would require that duplicate an entire support staff for that organization, replicating all the structures that already exist at the foundation like HR and Finance, insurance plans, etc. That would be a waste of money and time for no benefit. The better and easier approach is the one we have, which is WMF employees who work on the project with costs managed through a cost-sharing agreement. Re: compensation for any of the directors (technically, managers) of the LLC - they will never be compensated for that role, correct. The LLC exists as a legal structure but I don't think it makes sense to refer to "the LLC Staff": We're all normal WMF staff who happen to be working on this project. For example, in my personal case for the last year I've been part-time with Enterprise and part time with WikiCite.

2) Correct, that's what the LLC pays for when not compensating the Foundation directly for the work Foundation employees are doing on behalf of the LLC.

3) Correct. Also, the LLC as a single member wholly owned LLC is also subject to the reporting requirements of the Foundation.

LWyatt (WMF) (talk) 12:39, 31 October 2021 (UTC)

Thanks, Liam, sounds good. Good luck! Regards, --Andreas JN466 16:05, 2 November 2021 (UTC)

Clause about termination of contract if license is not respected?

I'd like to propose that any contracts should have a clause that lead to the termination of the contract if the license of the content is not respected. With respected, I mean not only perhaps technically being compliant, but being a role model and showing how good attribution should look like. ♥Ainali ^talk_{contributions} 18:56, 16 November 2021 (UTC)

License compliance, including attribution, is indeed part of the contractual obligations for any/all customers, yes. How attribution is formatted is, nonetheless, contextually dependent. Equally though, it is important to note that Wikimedia Enterprise contracts are not creating extra copyright-like restrictions via contract law which would make the content more restricted in its ability to be reused. Also, as an extra point of context: sometimes a search engine will display content that *could* come from Wikimedia sites but could also come from another of that organisation’s commercially-licensed data providers, so the absence of a Wikimedia attribution in any given specific context doesn’t necessarily mean that that the license is not being respected. LWyatt (WMF) (talk) 09:16, 17 November 2021 (UTC)

Sure, but I think we should not conflate providing sources with attribution. They serve different purposes, where one cater slightly more towards the reader's need and the other slightly more towards the editor's demand. ♥Ainali ^talk_{contributions} 20:47, 18 November 2021 (UTC)

Usecases of the Customers

There are currently the licensing discussions about the license for Wikifunctions and Abstract Wikipedia. Through that the discussion is also about how far will the content be used in a commercial way and how far is it possible to understand where the information comes from as an consumer, for example in an infobox in a search engine result or as a answer out of a smart speaker. I am interested in understanding how content is processed to get to the result the end user sees and for what the content is used in products or applications of the customers and I am interested in meeting customers of the Wikimedia Enterprise API. Do you think it is possible to organize a video conference where the customers can particpate, present for what they use the content and give the possibility to ask them questions and tell concerns people have about the way they use the data or content of the Wikimedia projects. I am interested in participating in such an event.--Hogü-456 (talk) 18:47, 25 November 2021 (UTC)

Dear @Hogü-456. I think the idea of providing a forum (perhaps a video meeting, or maybe in the future even a kind of 'panel session' at a Wikimedia conference) where some of the 'customers' of Enterprise's APIs can discuss what/how they use Wikimedia content in their services - would be very interesting! However, because this API service is currently very new, we do not [yet] have any "real world" case studies to show and discuss. Moreover, the larger the company, the slower it is to change their existing infrastructure and methods to the use of Enterprise APIs in their products. Over the next six months to a year we intend to be creating a section of enterprise.wikimedia.com which will highlight "case studies" and examples from customers who are happy and willing to tell the public what the are doing. As I mentioned in an earlier comment on this page we do not [yet] have our own independent social media/publicity profile (except for the website itself) and part of the process of building that presence is deciding what is the best way to showcase the cool and diverse ways that this service is, and can be, used. Video conferences with representatives of current customer organisations is indeed an interesting suggestion - we'll see if we can do it! LWyatt (WMF) (talk) 11:42, 26 November 2021 (UTC)

Revenue goal

What is the revenue goal of Wikimedia Enterprise for the time of the fiscal year 2022-2023 and what profit do you expect. In what class of taxation is the Wikimedia LLC classified and how high are the expected operating costs for the next fiscal year. After it is a buisness it is from my point of view important to treat the Wikimedia LLC like a company that is for profit in the buisness. The sentence with the plan about the taxation that I have found in the Wikimedia LLC Operating Agreement is not clear to me. I am not sure if I understand this sentence in the right way. I suggest to calculate the prices for the products you offer with a not to high markup. I suggest as a goal to reach a 5 percent profit with the company as the so called EBT, also known as Equity before Taxes. This is a from my point of view acceptable profit if it is made sure that the money is reinvested and a part goes back to the Wikimedia Foundation and goes there to the communty. If it is possible I think the Wikimedia LLC is a chance to try out how free knowledge can be used in companies and what kind of transparency can be offered to the customers. So if the amount the customer pays is lower or higher to reach the before calculated profit I think the customer should be informed. In my ideal understanding of business the prices should cover the costs but be not much higher or lower than that. There I see a chance to try if something like that can work in a company. I am interested in an office hour to hear about the current status of Wikimedia Enerprise and to talk about the plan for next year.--Hogü-456 (talk) 19:11, 30 April 2022 (UTC)

Dear Hogü-456,

With regards to the revenue goal: the “how much money will this raise” question on the FAQ has this note: As per the project's financial goals that were initially defined during the development-phase, the 2021-22 Annual Plan predicts "$10.2 million in contractual revenue and approximately $3.6 million in expense for Wikimedia Enterprise...". As this sentence implies - this is the assumption that was made early in the development of the project. Now that we are a further year ahead, things are becoming clearer - and consequently so is potential customer base. The most important thing is that we generate, as you said yourself, sufficient revenue such that Enterprise’s activities are self-funding - but not too much as to alter our fundamental focus on small-donations. Next month will also be the next public milestone and associated announcement, which will have details that answer some of your questions in this topic. ​​As per the project’s published Principles, we will also be publishing overall revenues and expenses at least annually - but that won’t be until there’s approximately a year’s worth of financial data to report upon. As things stand, we’ve only been “open for business” since January. The publication of the first full year’s worth of revenue/expenses will include more practical information, rather than the necessarily-hypothetical response I can give today.

As for the taxation topic: As also noted on that FAQ, “The assessment of appropriate tax treatment of the LLC activities has been coordinated with the Wikimedia Foundation auditors KPMG.” We expect that there will be a small portion of Wikimedia Enterprise’s revenues which are subject to unrelated business income tax [UBIT], while the majority will be tax exempt.

With regards to the pricing itself - we will have more information coming soon. There will be a pricing structure that the potential customers (and the interested public/wikimedians) can calculate for themselves. This is both ‘better for business’, and also more equitable + more transparent.

I have scheduled a community-office-hours for early June, where you can bring follow-up questions. See the details at "office hours" subsection below. Sincerely, LWyatt (WMF) (talk) 21:08, 11 May 2022 (UTC)

Thank you @LWyatt (WMF): for the answer,

currently I have not a clear description why I havent read your answer earlier as I was since then on the page and wrote something below. I think that the expected profit is much to high. I think for a business more than 25 percent profit in total are not responsible and so I think the prices for the services should be reduced and I think for the business the Wikimedia Foundation should pay taxes. Now I hope that there will be then discussions about it at the Office Hour or before during the annual plan discussions at this talk page and it is an example that I do not look to all relevant sides and will need to look more detailed and recently to board resolutions and talk pages. Because the business started already as you wrote there are some customers there.--Hogü-456 (talk) 20:12, 19 May 2022 (UTC)

How are we doing?

Hi @LWyatt (WMF):, Hogu asks above about revenue and an office hours (and so I am a +1 to both those queries) - I just thought I'd ask more generally about how Enterprise is doing? What progress has it made at standing up? Are the initial customers happy with the offering (if we are active)? Does Enterprise specify its year's plans at the same time as the rest of the WMF is doing the Annual Plan - if so, what are the key highlights? Nosebagbear (talk) 14:40, 9 May 2022 (UTC)

Hello Nosebagbear - I'm working on a reply to Hogu's questions so you can follow that thread. As for initial customers, highlights, annual plan... "watch this space!" While this page has been quiet in recent months, that is deceptive: a lot of work has been going on in the background and we're about one month away from our next major press-release with updates. I'll schedule and publicise a community-office-hours for that 'announcement week' where you can bring these kinds of questions (if they're not already answered by the news announcement itself). LWyatt (WMF) (talk) 10:43, 10 May 2022 (UTC)

Sounds good, look forward to the update Nosebagbear (talk) 11:19, 10 May 2022 (UTC)

Nosebagbear - This is now scheduled. See Talk:Wikimedia_Enterprise#June_9_-_"office_hours"_conversation. LWyatt (WMF) (talk) 21:43, 11 May 2022 (UTC)

Speaking for myself, I generally prefer having questions asked and answered on-wiki (as you have been doing admirably well, thank you). Office hours take a lot of time out of work/family life, the timing is always inconvenient for some people, and the information ends up dispersed across multiple pages, with no one place tracking the progress of discussions.

With this in mind, may I suggest you provide a summary here as well? Best, Andreas JN466 10:33, 26 May 2022 (UTC)

A video conversation is an addition to, not a replacement for, this talkpage. Some people prefer different methods of communication and so we’re trying to suit all preferences. If all participants are happy with it, we will record the video and upload it to commons and embed it here - as we did with the other ‘office hours’ calls last year. LWyatt (WMF) (talk) 11:23, 26 May 2022 (UTC)

LWyatt is indeed responsive - I think Andreas' request is specifically for a textual update/summary (along with any video upload to commons) Nosebagbear (talk) 10:17, 27 May 2022 (UTC)

I can't promise to be 'taking detailed minutes' of the conversation. However, if I've the presence of mind, while also facilitating the call, I'll try to take note of the approximate topic (and, if possible, the timecode to make it easier to 'skip to the relevant part') and put that in the file description when I upload it to Commons. LWyatt (WMF) (talk) 10:30, 27 May 2022 (UTC)

@LWyatt (WMF): YouTube videos these days have an automated transcript (available both with or without timecodes, accessible via the three-dot icon in the line under the video's title). Could you copy that out and put it onto the talkpage here, "hatted" (i.e. so it is only visible when a person clicks on it)? Andreas JN466 21:26, 8 June 2022 (UTC)

As with previous video-meetings, this will be conducted on Zoom (not Meet/YouTube). This allows for better moderation tools and video recording extraction/transcoding to Commons. We’ll see what the automatic transcription functionality is like (accuracy + extraction). LWyatt (WMF) (talk) 21:58, 8 June 2022 (UTC)

@Nosebagbear and Hogü-456: According to Advancement's quarterly review, Wikimedia Enterprise is behind schedule: "Although we successfully closed an initial set of paying customers for Wikimedia Enterprise at the end of the last calendar year, and have continued to have ongoing sales conversations with additional potential customers, we have been unable to close additional customers as quickly as we projected due to unanticipated legal and product requirements, and will not hit the revenue target for FY21/22." At the end of Q3, WE had achieved 30% of its revenue goal, unchanged from Q2. Progress has been made in Q3 though in terms of laying the groundwork for supporting small and medium enterprises. (On a different topic, the quarterly review also mentions that a number of Google staffers will join the Abstract Wikipedia team.) Andreas JN466 15:38, 15 June 2022 (UTC)

I wouldn't summarise that quarterly review description as "Behind schedule" but rather that it is "lower than the initial estimates". This might seem merely semantics but the difference is that the former implies there's a rigid 'sales expectation calendar' - which there isn't. Rather, the latter phrase indicates that as things progress, initial assumptions are steadily being replaced with real-world data. As I already stated above in the #Revenue goal subheading, and will reiterate here:

As per the project's financial goals that were initially defined during the development-phase, the 2021-22 Annual Plan predicts "$10.2 million in contractual revenue and approximately $3.6 million in expense for Wikimedia Enterprise...". As this sentence implies - this is the assumption that was made early in the development of the project. Now that we are a further year ahead, things are becoming clearer - and consequently so is potential customer base. The most important thing is that we generate, as you said yourself, sufficient revenue such that Enterprise’s activities are self-funding - but not too much as to alter our fundamental focus on small-donations... -- Wittylama (talk) 16:03, 15 June 2022 (UTC)

June 23 - "office hours" conversation

Come and meet the Wikimedia Enterprise team and ask questions you may have:

When & where: Thursday 23 June @ 1700 UTC on Zoom.

It's been a while since we last did one of these, and the intention is to have the next major updates about the project published earlier in that week, so this is a good opportunity to meet. Sincerely, on behalf of the team, LWyatt (WMF) (talk) 21:42, 11 May 2022 (UTC)

Hello @LWyatt (WMF):, in this year June 9 is a Thursday and not an Tuesday. Can you please correct the day in the link or tell if it will happen at another date. I am interested in attending and like that you have scheduled an Office Hour.--Hogü-456 (talk) 20:45, 15 May 2022 (UTC)

Hogü-456 thank you very much for noticing and informing me. I have corrected it here, and in the main page. It is indeed THURSDAY June 9. LWyatt (WMF) (talk) 21:02, 15 May 2022 (UTC)

NOTE: This event was originally advertised as on the 9th. It has now been delayed by precisely two weeks to Thursday 23rd. I have updated the sections subheading to ensure it is not confusing. (Ping Hogü-456 in particular).
The reason for this change is because the major update announcement itself (mentioned above) is being pushed by two weeks for a couple of pragmatic reasons including covid. Sincerely LWyatt (WMF) (talk) 11:16, 2 June 2022 (UTC)

Hello @LWyatt (WMF):, I have changed the date at the page belonging to this discussion page. There was the 9th of June mentioned as the date, when the office hour will happen. Can you please mark the page for translation or where can I ask for that.--Hogü-456 (talk) 19:05, 9 June 2022 (UTC)

Ah - thank you Hogü. I thought I had updated them all, but missed that (prominent) one! I will do. LWyatt (WMF) (talk) 19:08, 9 June 2022 (UTC)