Google’s founders didn’t anticipate that the search engine built on PageRank would one day earn billions of dollars via advertising revenues. Google was founded as a technical achievement with superb market positioning — at the nexus of information demand and supply. It was free and useful, becoming popular quickly. From there, other clever people came on, and have been able to leverage this positioning into the multi-billion-dollar technology and data company now known as Alphabet.

Of the two fundamentally important aspects of business — positioning and execution — it’s arguable that positioning is by far the most important.

We now see some interesting positioning moves by entities with plugins designed to exploit Green OA or address access issues in a virtual, seemingly decentralized way — Unpaywall, Kopernio, and Anywhere Access. There is also the STM community’s effort to improve access, RA21. As a group, these and a few others coming onto the market represent commercial responses to the presence of Sci-Hub, ResearchGate, and Academia.edu. While they have interesting positioning, their execution differs in some key ways.

Kopernio was created by a serial entrepreneur, Jan Reichelt, who co-founded Mendeley. Earlier this year, Kopernio was acquired by Clarivate Analytics for an undisclosed sum, making this two acquisitions for Reichelt in a relatively short period of time. He clearly knows how to create companies others want to buy. (Interestingly, Unpaywall has a licensing deal with Clarivate via Clarivate’s Web of Science. Whether this will continue much longer isn’t clear.)

When Kopernio connects users with articles on a publisher’s site, the articles are placed in the Kopernio PDF frame, eliminating the publisher’s major branding elements. To do this for paywalled content, Kopernio has to acquire the user’s subscription entitlements — if the user is on-campus, the IPs and login information are passed through Kopernio; if the user is off-campus, there is a one-time handshake needed to entitle the off-campus computer, which can include profiling the user’s computer as well as acquiring the access credentials, which are stored in the plugin.

Kopernio makes three promises when you install the plugin:

1. Turbo charge your literature search.
2. Jump over paywalls.
3. Work off campus.

Note that during this process, Kopernio and Clarivate are acquiring user credentials — these are stored in the plugin, and Kopernio asserts they are not transmitted to their servers, but things can change. The terms of use are with Kopernio. The Privacy Policy leads to a Clarivate page, which seems a broader catch-all for Clarivate overall and non-specific to Kopernio except when it states, “‘Clarivate Analytics’ includes all the businesses of the Clarivate group, including separate brands like CompuMark and Kopernio.” Both Kopernio and Clarivate are based in the UK, so GDPR rules apply.

Their privacy terms include this clear statement:

We collect personal data directly from you, for example, the contact information and log-in credentials that we use to provide our services to you, and the professional profile details that you choose to enter into our products.

Kopernio struck a deal in July 2018 with Caltech to conduct a pilot with 2,000 Caltech researchers, providing the researchers with one-click access to their entitled content “via the identification of a researcher’s subscription entitlements,” which is a very interesting phrase when it comes to the positioning of Kopernio in the scholarly publishing ecosystem. According to people who know more about this deal, Caltech is getting some of the data Kopernio/Clarivate is gathering during this pilot. An interview request with the relevant Caltech person was not granted in time for this post.

Given that the Kopernio plugin is acquiring not only individual information on users but also access credentials, it has an interesting position, especially as Clarivate is a data company through and through. Kopernio also smacks of the kind of company we’ve become accustomed to online — as Leonid Bershidsky wrote about recently in Bloomberg BusinessWeek:

This is a time when companies whose innovations are more intrusive than useful, more gimmicky than problem-solving, operate with business models that either burn investors’ cash or turn the users into products.

This article, which was about the writer’s reluctant admiration of Apple as “a rock of common sense, sobriety, dignified engineering supremacy, prudent financial and supply chain management, effective marketing, and customer-oriented retailing,” includes this quote from Apple CEO Tim Cook:

If you look at our model, if we can convince you to buy an iPhone or an iPad, we’ll make a little bit of money. You’re not our product.

Not that Kopernio is above using incentives to get more user information. Kopernio Premium is offered as an incentive to users who refer someone else to Kopernio, offering more storage and other features to users who spread the word.

Kopernio’s approach is reminiscent of the industry’s RA21 initiative to create seamless authentication, which recently resolved in a recommendation to pursue a similar browser-based approach to seamless access, but one which claims to preserve user privacy.

Kopernio’s user experience is pretty aggressive. Most plugins install in a sleeping state, and you have to activate them. Kopernio installs in an active state, and begins pinging you as soon as you visit a domain it thinks is relevant. It also can’t be shut off via the plugin once installed, with the user’s only option to go into the browser’s settings to turn it off, or to uninstall it if you don’t want it surveilling you. This approach creates a barrier (probably intentional) to deactivating the plugin.

Kopernio sources articles from controversial venues, including ResearchGate and Academia.edu. It may also use the Unpaywall database, but I couldn’t corroborate this speculation from some of the publishers I spoke with. The copyright controversies with ResearchGate and Academia.edu are well-established. Some publishers are considering blocking Kopernio if their practice of gathering PDFs from these sites doesn’t change. Given how important it is to Clarivate that Web of Science continue to have access to publisher sites and publisher goodwill, it seems these could short conversations. However, in speaking with publishers who have interacted with the leadership of Kopernio, there is a stubborn refusal to modify their approaches, some of which (see Sidebar #1) are questionable on their face.

Unpaywall is run by ImpactStory, a non-profit with stated revenues of $15,058 in 2015 (according to their 990) against expenses of $215,900, so the recently announced grant of $850,000 from the Arcadia Fund matters to their viability. Whether the team’s aim is a long-term one remains unclear, as the reputation of Heather Piwowar and Jason Priem in the market is one of cleverly changing the game as happy warriors. They are another set of serial entrepreneurs, although less financially successful than those behind Kopernio.

Technically, Unpaywall uses relatively brute force tactics to identify Green OA content. With a database they claim approaches 20 million articles, Unpaywall matches the DOI or URI with their database, and when the plugin is activated, Unpaywall will tell users when and where an article can be had for free. The Unpaywall plugin also installs in an activated state, meaning it is in surveillance mode from the moment you install it. However, you can turn Unpaywall off without uninstalling it.

Sidebar #1: As a matter of web manners, Kopernio is pushy and unyielding. From what I can see, once you install Kopernio’s extension in Chrome, you cannot turn it off except by finding the proper browser settings and turning it off there, or by uninstalling it. Otherwise, it is always surveilling you. Unpaywall starts in an activated mode, but you can turn it off without going into third-tier browser menus or uninstalling it. I believe both are overstepping an important boundary by installing activated, but Kopernio goes into my obnoxious zone with its plugin’s practices. 

Ostensibly, Unpaywall represents a great way for publishers who have strong Green OA policies to let users take advantage of these free-to-read articles. After all, publishers are the main providers of Green OA. However, Unpaywall may create the impression that technology has enabled article access, rather than policies, leaving publishers once again without the credit they deserve for liberal access policies. Also, the threat of Green OA has always been blunted by having a disorganized mess of free articles tucked away in unpredictable places, hindering discovery and usage. Unpaywall is organizing this mess, so liberal access policies that once were adopted with little risk might need to be reconsidered.

Unpaywall’s privacy policy is also an interesting read, as they divulge that they acquire personally identifiable information (including the IPs used), and also acknowledge that even if you activate Do Not Track in your browser, they continue to track you (“At this time, ImpactStory does not respond to or alter its practices when it receives a Do Not Track signal from a user’s web browser”). (Note: After publication, Jason Priem commented that their policies are unclear, and that they will be modifying them.) There is no mention of GDPR in their privacy policy (the company is based in the US, in North Carolina). In July 2018, Elsevier announced a deal to integrate Unpaywall into Scopus. It’s unclear if this will be a lasting relationship, or if Elsevier is dancing with Unpaywall to learn its moves before it strikes out on its own. Unpaywall’s founders are devoted to open source solutions, so Elsevier may just port the Unpaywall code at some point in the not-too-distant future. Unpaywall appears to be funding itself via licensing fees around its dataset, as well as via the aforementioned grant.

If Unpaywall is widely adopted, it may portend a shift away from Gold OA, especially if embargoes as practiced now are truly tolerable to non-paying users. But the nagging question will be around the long-term value of a Green OA discovery database, especially one the founders claim is “open.” The founders may only want to shift the market, and may not have long-term commercial goals with Unpaywall. One could imagine this becoming an infrastructure element like the CHORUS API, funded at a far lower but sufficient level by publisher members or a similar community approach. Or Unpaywall may be the first of many such services, and not the best. Just this week, Third Iron introduced LibKey, a very similar PDF finder for Green OA content. Soon, the market may be flooded with these, which means the differentiation will come from marketing and sales, a situation where larger organizations will again prove their positional advantages.

Sidebar #2: Given the two players striking deals around the two most prominent plugins, my speculation that Elsevier and Clarivate are squaring off in the market has been reinforced.

Anywhere Access comes from Digital Science, a property of the Holtzbrinck global conglomerate. Anywhere Access is reminiscent of ReadCube, a product that had trouble moving beyond a few big publishers for a variety of reasons, with a main one being the fact that caching PDFs for optimized delivery is extremely expensive at scale (yes, digital is expensive). To circumvent this problem, Digital Science has taken the browser plugin approach to decipher access rights for the user and present a clickable PDF button with some enhancements to the PDF — light annotation, export options of reference managers, and so forth. Anywhere Access is partially a library play, generating data for libraries around usage, usage problems, and so forth. Anywhere Access doesn’t store PDFs, and promises publishers abilities to price a la carte access, improve user experience, and so forth. Like the others, Anywhere Access gathers a lot of personal information from its users, and users agree to let their information be used (in an aggregated, depersonalized way) by advertising entities.

From an access perspective, we have a few different approaches on the field to consider:

• Kopernio, which gathers user credentials for a large multi-national company, and which promises to help users “jump paywalls”
• Unpaywall, which gathers personally identifiable information for a startup working with a large multi-national company, but limited to Green OA content
• Anywhere Access, which uses library information to present users with a clear, single-click path to the content they have the rights to read via their institutional subscriptions, while offering publishers promises of more usage and pricing controls
• RA21, a collaborative, non-commercial initiative built with input from the community, which is designed to preserve user privacy, and will work for publishers and libraries of all sizes

Of course, guess which one — the three commercial offerings associated with huge companies, or the one community-collaboration-based — is distrusted and disliked the most, especially by librarians?

If you assume an anti-publisher sentiment, this will help you triangulate on the answer. A startup that finds Green OA content publishers make available, gathers user data, and charges licensing fees? No apparent issue. A large non-publisher corporation using a recent acquisition’s plugin to store end-user data including access credentials to multiple institutions? Librarians apparently are quite happy with Kopernio. A digital-native for-profit start-up associated with a huge global information business and running products that gather reams of user, commercial, citation, and library information? No major controversy yet. A free, privacy-protecting access option created with input from multiple stakeholders and developed by a coalition of publishers? Yes, you guessed it, the last one is the answer. Because it’s associated with publishers, RA21 has faced skepticism, distrust, and disdain from the start.

All of these services are predicated on the subscription model being in place in a substantial way.

There is a common thread to these services — all are predicated on the subscription model being in place in a substantial way. Unpaywall has that assumption in its name, while Kopernio only matters if access credentials need to be managed. Anywhere Access’ core functionality and suite of ride-along services (IP recognition, usage reporting, access errors) only really matter if site licenses exist. Of course, RA21 is explicitly designed to make access to subscription content easier for users, with the inciting moment being the emergence of Sci-Hub — which raises the question, with Sci-Hub still operating, do any of these matter? More to the point, what if someone were to make a plugin as simple and effective as Unpaywall or Kopernio or LibKey or Anywhere Access, and point this at Sci-Hub? Would users defect to this? Would libraries approve? Would publishers sue? The answer to all three may be affirmative. The consequences would be more time spinning our wheels in the mud of uncertainty.

The value of content is not in dispute if you read between the lines. It’s just a matter of how to commercialize that value. Traditional publishers want to take the Apple approach, and sell it directly to the user or their purchasing agent. To make this model more effective, they’ve created RA21, and are rolling it out later this year. Start-ups like Kopernio are taking a different approach, seeking to commercialize access by gleaning access credentials and personal information, mixing the appropriate parts with other corporate offerings at Clarivate, and ratcheting up their service levels and revenues, making the users into the products to some extent. This is the Facebook approach. Then we have Unpaywall, which is a combination of the two, gleaning data from users while licensing its database in a traditional manner, which is akin to the Linux model, complete with a desire to upset the status quo just because they can. Finally, Anywhere Access is seeking to sell to both sides of the market — buyers and sellers — with an offering they claim benefits to both. This is the Amazon approach.

The tensions among these approaches are familiar — privacy, piracy, copyright, threats to publishers, opportunities, user-satisfaction, user exploitation. We’ve seen all this before in various guises with Facebook, Twitter, Amazon, and Google.

Google, mentioned at the outset as having assumed a position relative to content which some feel devalued it (and which definitely commoditized it to some degree), has learned a valuable lesson from some of its experiments — that the value of content is directly related to the value of Google, as noted in a recent Digiday podcast interview with Richard Gingras, VP of News at Google:

The relevance and value of Google search for billions of users is based on the concurrent depth and breadth of the knowledge ecosystem of the web, to the extent it deteriorates, it’s not good for Google search.

The surprising commonality here is that the ultimate value of Kopernio, Unpaywall, Anywhere Access, and RA21 depends on a healthy subscription model. A complete flip to OA means all of these become largely irrelevant, so by extension (pun intended) they are betting against Gold OA. Each is pursuing value aligned with the subscription value of content.

Given the history here, I think we’ll find that of the set, Kopernio will have the most trouble and have to handle the most controversy in the long-term, while RA21 will continue to face the anti-publisher bias that has been cultivated for years in our ecosystem. Anywhere Access has a harder row to hoe, if only because Digital Science has so many offerings, a ReadCube base to migrate, and sales uncertainty to face in two markets. This leaves Unpaywall with the smoothest path forward. But they also have the most uncertain and uninteresting destination, a ride that may be reminiscent of Thelma & Louise. However, given Unpaywall’s founders’ apparent discomfort with or disdain for commercial success, they may just be in it to enjoy the ride.

(Thanks to multiple people — especially AC, MC, and JE — for a great deal of help with this post.)

(Note: An error in my understanding of how the Kopernio plugin works led me to believe user credentials were being stored on the server side of the service. I’ve been reassured by Clarivate that all credentials are encrypted locally in the browser. The post has been modified accordingly. My apologies for the mistake.)