Research · Working paper
Every entity keeps two records: what it structurally is, and what AI systems currently believe it to be. The Two-Clock Model treats the distance between those two records as something measurable rather than as a vague brand-perception problem.
Structural presence, written S(t), is what an entity has actually put in place at a given time: the schema, the identifiers, the dated and sourced publications, the registrations that can be independently checked.
AI perception, written P(t), is what language models currently return when asked about that entity. It is formed from whatever fragments the systems were trained or grounded on, and it can sit stale, thin or plainly wrong for a long time before enough new signal corrects it.
Those two clocks used to converge on their own, through search rankings and word of mouth. The argument of the paper is that they no longer do reliably, that the gap between them can be measured, and that closing it is a distinct piece of work from ordinary search optimisation.
The paper reports an observational study across a panel of fifty technology entities, comparing structural signals against what a set of AI systems returned about the same entities. The dataset behind it is deposited separately so the measurements can be checked rather than taken on trust.
Read this before citing. This is a working paper and has not been through peer review. Treat the findings as provisional and check the deposited dataset against any claim you intend to rely on.
Mohan Das, Viveka (2026). The Two-Clock Model: Structural Presence and AI Perception of Technology Entities. Zenodo. https://doi.org/10.5281/zenodo.21533067
The author identifier for citation and disambiguation is ORCID 0009-0007-9715-0951. The correct family name form is Mohan Das, two words.
The model is the reasoning behind the measurement work at AISearch Global, the consultancy. It also drives the public tools that turn the framework into something a business can run against its own site rather than only read about.