Thread by @sanguit

The limits of applying the 'platform argument' to BigTech firms, particularly Facebook, Google, Twitter etc.

A thread. 👇

The common narrative today - and also in a lot of my work, including Platform Revolution - frames Facebook, Google, and Twitter as multi-sided platforms.

Are #Facebook, #Google, #Twitter multi-sided?

Absolutely.

But viewing them solely with that lens has limitations, which interferes with our efforts to design regulation.

What really matters is the unit of analysis.

Is the unit of value (the core value unit, in my work) content?

Or is the core value unit attention?

From a 'product management' perspective, if you claim you're building for users, the core value unit is content.

More on core value unit from my 2013 article: platformthinkinglabs.com/materials/the-…

But from a 'business model' perspective (unit of value, a portion of which is captured), the core value unit is attention.

If the core value unit is attention, these companies are very much traditional pipeline businesses harvesting attention, packaging them into data and targeting products, and selling them - literally - to the highest bidder.

On Pipelines ve platforms: hbr.org/2016/04/pipeli…

Why is viewing attention as the core value unit important?

Because business models are built around scarce and tradable resources. Tradability affords value transfer and scarcity affords negotiation leverage in such transfer.

Attention is both scarce and tradable.

Attention is scarce - yes, because the sum total of human attention is a finite number.

But more important, because with an abundance of content and interactions, our decision-making ability gets depleted, allowing our attention to be more manipulable as a resource.

Attention is also tradable. Sophisticated auction mechanisms trade our attention (and the attendant decision - which is already manipulable) to the highest bidder.

Hence, to regulate these companies, we really need to take off the 'platform' hat we don for product management and put on the 'pipeline' hat, viewing these companies as pipelines of attention.

A further issue here is one of 'agency'.

Pull (search) involves some agency.

Push (news feeds, recommended videos on Youtube) involve much lower agency.

Push models can be (and are) fine-tuned to harvest attention further.

Shifting the unit of analysis from content to information, our notion of production also changes.
'Producers' on these 'platforms' are producing content but the consumers of content are 'producing' attention which may be transformed, packaged, and traded at scale.

Another lens to figure the value of analysis is competition.

When products are substitutes, they compete in the same market.

Arguably, an Instagram photo is not a substitute for a Whatsapp message. But the attention harvested on the two 'platforms' are arguably substitutes.

Facebook acquires Instagram and Whatsapp not because it wants to increase its 'content' portfolio but because it wants to minimize competition from substitutes to its 'core product', which is attention.

In 2021, this should be obvious.

Facebook often uses the term 'platform' to skirt regulation. But regulators need to take off the 'platform' hat and see these companies simply as harvesters of attention.

Another issue that distracts regulation is focusing on data rather than on attention. Eventually, what is traded is attention (and the attendant action, which as mentioned earlier is manipulable). Data is merely a mechanism to manipulate said action.

Regulators today regulate data as the scarce and tradable resource. Hence, regulations like GDPR focus on ownership and access of data as a scarce and tradable resource.

Instead, attention and the attendant, manipulable action should be the focus of regulation.

Practically, this calls for new regulation of algorithmic advertising rather than regulation of data ownership and access.

Antitrust in itself is limited in regulating algorithmic advertising for a variety of reasons.

First, antitrust view competition as a solution to concentration. Unless we see attention as the resource being competed for, the definition of competition and hence the regulation can be obfuscated.

Second, more competition - while better than concentration - is not a sufficient response. This is because AdTech is modular and over time, lower layers of AdTech get commoditised and standardized across the industry. This results in tacit collusion.

For example, if multiple companies use the same adtech at the backend, more competition on the service frontend won't offer any solution to the problem of attention harvesting.

Third, regulation needs to work across the stack - from
(1) regulating frontend services which use 'dark UX patterns' to manipulate actions, to
(2) 'mid-end' ad-tech that matches attention to the highest bidder, to
(3) back-end data markets that commoditize data

These are all hard problems. While we're far from perfect solutions, we need to at least start framing the problems right.

1. View attention - the scarce and tradable asset - as the unit of analysis
2. Understand that neither breaking up nor competition will solve the problem
3. Shift regulation from data to algorithmic advertising
4. Regulate across the adtech stack

My next post will build on some of these ideas at platforms.substack.com