Thread by @jameswstaylor

Why, specifically what we are building at @ParticularAud, will enable the second pillar @Facebook
/ thread

tech.fb.com/inside-faceboo…

Quote ~ “The second pillar is the use of AI, context, and personalization to scope the effects of your input actions to your needs at any given moment."

@ParticularAud is a cross-platform behavior network developing a context aware + internet scale filter for information. This allows contextually relevant information to be recommended.

~ "it will require building powerful AI models that can make deep inferences about what information you might need or things you might want to do in various contexts, based on an understanding of you and your surroundings, and that can present you with the right set of choices."

Context is inherent in recommendation networks, context is the item you view from which further items can be recommended. Or a sequence of items for that matter.

Context may be a URL, or an element within that URL.

Like images or a descriptive term... or any other 'thing' or interaction.

Context could be a set a coordinates, a time of the day, or a stage in a sequence of activity.

Mapping context to outcomes is how Spotify continues to play you music you love after a playlist has finished; and how your Netflix account is completely unique to your tastes.

Facebook's rapid expansion of tracking tags and SSO across the web will help them to map information across the internet.

...and it goes without saying - they have timestamped, geo-mapped, and even conversational context from across their many apps you may or may not have on your mobile phone.

Specific domains take tremendous specialism, however. We've already built this specialism out for the commerce domain. Tracking items down to a SKU level, with semantic understanding is a challenge that we have already overcome.

Whether Facebook intends to do this themselves, or open up an app store as we have witnessed on practically all other consumer devices remains to be seen.

What we do know is that app developers have driven exponential improvements on our consumer devices of today. App developers deliver untamed and infinite innovation potential.

The number one problem these next gen devices will incur is a filtering problem. You won't have the screen real estate (or time) to compare results in multiple tabs.

@benthompson on aggregation theory attests to the benefit of providing the best filtering experience. It is how super aggregators (such as @Google) are winning the consumer internet today.

It is how @netflix and @Spotify are winning in their specific domains.

@ParticularAud we are consolidating the entire experiential layer on the internet by creating the👏best👏filtering👏experience.

The context-aware, automated, and intuitive filtering experience we expect from platforms like Spotify and Netflix is being made accessible via API

Meaning any device will be able to access our internet-scale network.

Networks solve filtering problems by defining relationships between things

@reidhoffman has long known that networks compound the value of its constituents

enabling their contextually relevant discovery

it is a shame (in the west) that equivalent internet-scale networks have resulted /pretty much entirely/ in advertising business models

@conniechan has long been singing praise to the commerce-first foundation underpinning China's consumer internet leaders

it is fair to say, an advertising-first internet is a sub-par experience - after all, ad relevancy is limited to the available ad supply at that time

...and who loves ads?!

this is why we started @ParticularAud in the commerce domain. The commerce domain is the most valuable network but is fragmented across ∞ single platforms

the winning network will be cross-platform with the greatest scale

so we started building that network

GREAT businesses can be built off this network (starting with our low-code API + no-code SaaS offerings)

but the future of how we interact with digital commerce depends on it

...and once we're done we'll move on to any other contextually relevant information you might need

@threadreaderapp unroll