CreaComp

Competency 2.1

Understand recommendation algorithms

Boundary this domain concerns the understanding of systems, never their productive use. Understanding an algorithm belongs to Analyse; publishing by exploiting its logics belongs to Make. Understanding AI belongs to Analyse; creating with AI belongs to Make.

Definition

Understanding what recommender systems optimise, the signals they read, how they distribute content, and their effects on individuals as well as on the visibility of creative work. This competency grounds offensive algorithmic literacy: not “how do I break through”, but “what is happening under the bonnet, and what do I choose to do about it”.

Components

  • The notion of signal: watch time, early engagement, completion, recency, regularity
  • Personalisation: why two people do not see the same world; bubbles and reinforcement loops
  • The distribution of a piece of content: initial test, audience tiers, effects of native format
  • Differences in logic between platforms (social graph, interest graph, search)
  • Individual and societal effects: attention capture, polarisation, homogenisation of formats

The four levels

  1. N1 Discovery Lower secondary · complete beginner

    Is able to explain in their own words why two people do not see the same feed, and to recognise that a piece of content was recommended to them rather than chosen.

    Evidence of assessment

    Compare and explain the feeds of two different accounts on the same platform.

  2. N2 Application Upper secondary · advanced beginner

    Is able to name the main signals read by an algorithm and to deliberately adjust their consumption or publishing accordingly.

    Evidence of assessment

    Deliberately alter their recommendations on a platform and document the approach and its effect.

  3. N3 Adaptation Tertiary · junior professional

    Is able to analyse the distribution of their own content from real metrics, to formulate a hypothesis for improvement and to test it.

    Evidence of assessment

    Publish two versions of the same content with different distribution strategies and analyse the performance gap at day 7.

  4. N4 Orchestration Experienced professional

    Is able to compare the recommendation logics of several platforms, to anticipate the effect of an algorithm change on a strategy, and to turn that into distribution decisions.

    Evidence of assessment

    Produce a multi-platform algorithmic audit, the distribution decisions drawn from it, and their measured effect.