Web Infrastructure Search Endowment (WISE)

The Ad Hoc Levy on Defaults (AHLD) is a system in which a portion of search engine revenue is levied and used to pay browsers. (It is also used for operating systems and other agents, but our focus in this document is browsers.) The levy ensures that browsers, which constitute criticial infrastructure for the web, are provided free of charge to people. It can take multiple typical forms:

• Default placement: the browser vendor sells its default search function to a search engine. Defaults have an outsized impact on search volume as few users change them. This practice also includes selling a presence (that isn't the default position) on the list of search engine alternatives that browsers provide.
• Royalties: the browser vendor gets a share of the revenue when directing traffic to a given search engine.
• Intra-company transfers: the browser vendor and search engine vendor are the same company, and revenue from the latter contributes to paying for the former. In effect, this is a horizontally integrated variant of default placements that is cheaper and of greater strategic value. (For smaller search engines, it may also be the only deal they can make.)

These deals are negotiated bilaterally, and in practice an AHLD agreement between a browser vendor and a search engine vendor can marry aspects from several of these forms.

AHLD levies are economically significant and constitute one of the larger commercial exchanges on the web. For instance, in the year 2021 Google Search spent over $26 billion USD on default placement deals ([26-Billion-Default]) — which does not include its intra-company transfer costs — and paid 36% of its search advertising revenue made in Safari to Apple ([Apple-36]). In 2020, Mozilla made 86% of its revenue from the Google Search levy ([Mozilla-Revenue]). Note: We have to use relatively old numbers due to the unfortunate lack of transparency that the AHLD system suffers from. To the best of our knowledge, the situation has not significantly evolved since.

The AHLD levy was an excellent invention and remains a great idea twenty years on. Taking a system view of the web, search is indeed one of the logical places at which to apply a levy. Search extracts value from content and behavior on the web (it would have no value otherwise) and renders it available at a functional choke point (it is one of the required components of web discovery). In turn, the value of web content is only possible because browsers provide critical infrastructure services that render that content available and attractive to people, and for the most part safe. In an idealized view, we can envision a virtuous, regenerative cycle: the search levy finances core web infrastructure and that core web infrastructure makes the web successful such that search engines enjoy strong businesses that can be levied from.

Unfortunately, the AHLD levy has failed to evolve as the web grew from a time of relative infancy when only 14% of the world population (fewer than a billion, [Internet-Users]) used the Internet to the essential part of society that it is today. Over time, serious shortcomings with this ad hoc approach have surfaced that have not been addressed. It is incumbent on the web community to step up and hammer out an agreement for a better levy, as well as to commit to maintaining it so as to avoid the accumulation of problems ([rfc9413]). The changes need not be revolutionary since the core principle — a levy on search to pay for web infrastructure — can readily remain the same. However, as pressure mounts to address the problems caused by an unmaintained, ad hoc system, if we fail to act we may lose the levy altogether. Should that happen, the web will suffer from the impact on its infrastructure and we may find it very difficult to maintain a high-quality browser engine, let alone a diverse set of them.

The rest of this section captures the shortcomings of the AHLD levy.

Search is a key architectural component of the web, browsers provide critical infrastructure, but unfortunately there is no transparency into the system. Almost everything that the web community knows about AHLD we know thanks to material released via court cases. If we are to take seriously the W3C's goals of building a web for all humankind, we need to ensure that the web community is able to evaluate how the beating heart of web infrastructure operates.

There is also no accountability concerning the bilateral ad hoc deals made within AHLD. A search engine may impose additional requirements on browser vendors that might not be in users' best interests (e.g. that the browser must not implement certain privacy protections) without oversight. A system deployed at such a scale needs to be trustworthy.

It does not respect the priority of constituencies ([ethical-web-principles]). Search deals made via the levy can have terms that forbid the browser from intervening on the search engine results page (SERP) even when such changes would be beneficial to the user. For instance, a search engine may benefit from making its ads look like search results or from pushing users to be logged in, sacrificing privacy. A browser is expected to counteract such practices, but the AHLD levy prevents that.

The system gives power to search engines over how browsers work, including over aspects of browsers that may not seem search-related, for instance privacy features that can be used for advertising attribution may have to be approved by the search engine before they can ship in a browser. The lack of public accountability over the royalties part of the system means that there is a strong information asymmetry between the search engine and the browser in terms of where the royalties come from and what may affect them, which empowers the search engine to threaten potential revenue loss when the browser makes a change they dislike in ways that cannot be verified. Because of this, the search engine is both judge and party in the relationship, and in a position to exert undue influence over browsers.

Browsers are paid by the AHLD levy and may be subject to private requirements imposed by search engines, but from the perspective of the web community no requirements are placed on browsers in exchange for benefitting from the system.

The levy system pays for browsers but it does not pay for browser engines. Browser engines are the more complicated component and ought to be supported directly. As things stand, a browser can use an open source browser engine, collect funds from the levy, but not contribute anything back to the browser engine. This free riding is detrimental to the maintenance of a rich ecosystem of browser engines.

Worryingly, most appropriated funds don't go towards web infrastructure. While the purpose of the levy is manifestly to support web infrastructure, the overwhelming majority of funds is directed elsewhere. To take but one example: based on 2021 numbers, of the $26 billion USD that Google Search paid in levy, $18 billion went to Apple (about 70%). There is scant evidence that Apple, a publicly-traded company, spends in the vicinity of $18 billion USD per year on web infrastructure. The levy therefore suffers from very low efficiency, and in turn this causes value produced on the web to be directed outside of the web, which subsidizes its proprietary competitors.

The logic of applying a web infrastructure levy on search is because search can be considered to distill the value produced by publishers, ecommerce sites, and content creators. Unfortunately, publishers, ecommerce sites, and content creators have no say in a system eventually build on their work. While indexing content for the purpose of providing links back to the original source is evidently in everyone's interest, the lack of checks and balances in the system has led to a race to the bottom in which search engines provide fewer-and-fewer links back to sources ([AMP]) while increasing the number of non-linking purposes for which they process the indexed content, such as generative AI.

Last but not least, the manner in which the AHLD system is structured means that it mechanically increases concentration in the search market. Few people change their default search engine (particularly on mobile devices), which means that purchasing a default placement effectively purchases market share. In turn, the search engine with the highest market share has greater profits from which to pay a higher price for further default placements. This eventually leads to almost every browser defaulting to the same search engine, and that search engine dominating the web by insuperable margins. In turn, this artificially-crated and -sustained concentration creates further problems:

• Decreasing search quality. It is both much easier and far more valuable for sites to optimize for a single ranking function that changes occasionally than for many different ones that evolve in different directions. This makes signals of content quality easier to fake, thereby empowering low-quality sites to rate higher. The decrease in search quality ([Bad-Search]) and the effect of search monoculture on the web ([Perfect-Webpage]) are well-documented.
• Higher misinformation. As signals of quality lose their value, it becomes increasingly easier for misinformation to weave its way into search results.
• Loss of pluralism. A search engine defines an editorial policy: like newspapers or TV channels before it, it determines what is and isn't relevant. The modality differs in that the editorial policy is applied on demand and automatically, but it is no different from and no less subjective than that set by other forms of media. Having close to a single editorial policy for the entire web eliminates media pluralism, which has negative impacts on society and democracy, and leads to privatized censorship.
• One company pays for all browsers. Every major browser engine and browsers accounting for at least 90% of the market are all paid for by one single search engine. Without ascribing malice or asserting misbehavior, this level of privatized control over critical infrastructure is incompatible with a free and equal Internet.

While it is necessary to take stock of the AHLD's shortcomings, and while having left it to linger too long in outdated ad hoc arrangements has caused these shortcomings to fester, these problems, once identified, can be addressed. The web community has a responsibility to do so quickly and effectively, as well as durably.