Site status — July 2026: The news aggregator runs as normal, around the clock. The two research projects hosted here, Canary 2.0 and Fathom, are both complete. Their pages remain live as a record.
KnowEntry was built to solve a specific problem: keeping up with the volume and complexity of AI industry news without wading through cluttered, image-heavy news sites. The site has since done two jobs. The first is the aggregator, which runs continuously. The second is to publish research, and two projects have now run their course here.
The aggregator
The aggregator side is simple by design. Six curated sections, General AI, Academic Institutions, Big Tech, Technical and Research, AI Funding, and YouTube, pull from established publishers automatically, around the clock. No images, no advertising, no distractions.
This part of the site continues to run, and is unaffected by the closure of the two projects described below.
Canary 2.0: AI sector news sentiment
Canary 2.0 was a personal research project built and operated by Roger Smolski between February and June 2026. Each morning it collected between 800 and 1,000 news articles, scored them for sentiment and relevance using Anthropic’s Claude, and published a daily reading for six AI industry categories. Alongside it ran the Semantic Volatility Index, a composite measure built to detect a change in the character of coverage before that change reached the headline numbers. Roughly 55,000 articles were scored across the run.
The project registered a single hypothesis in advance: that a three-day change in a category’s sentiment score would predict the next day’s return for that category’s stock basket. It did not. No category cleared the significance threshold, and four of the six correlations came out slightly negative. In retrospect the hypothesis was testing the wrong outcome variable at the wrong timescale. The value of fixing it beforehand is that this could be said plainly afterwards, rather than quietly adjusted.
What did work was the Semantic Volatility Index. It detected and tracked two complete narrative rotations as they happened, and, in a third case, correctly overturned its own earlier reading when further data showed the shift to be cyclical rather than structural. The index is a genuine detector of narrative change in sector news coverage. It was never a price forecasting tool, and Canary was never a trading or investment signal.
Canary 2.0 concluded on 4 June 2026. The dashboard and methodology pages remain live. The full account is on the Project Conclusion and Final Report.
Fathom: processing complexity
Fathom, completed in July 2026, is a project of a different kind. It set out to test the central claim of the essay Consciousness and Complexity by Roger Smolski: that consciousness is processing complexity, meaning how far a system draws its own past together into its present, and that it comes in degrees, with no sharp break in nature. The essay states plainly that the tool needed to settle this is missing. Fathom was an attempt to build it.
The instrument, ΔAIS, asks how much a system’s recent past predicts its present, once you subtract what a system with the same surface statistics but no genuine memory would produce. It was calibrated against systems whose answers are already settled, and then aimed at the case that mattered most: a language model watched across its whole training, from a random starting state to a competent one. On that system the measure went quiet. It could not certify genuine integration in the trained model at all.
The theory was therefore neither confirmed nor refuted, and stands where it stood. What Fathom established is narrower and, in its way, more useful: where the current means of measuring the essay’s quantity stop working, and why a measure that reads a system’s parts one at a time is the wrong instrument for a system whose work is done by its parts acting together.
The findings are set out on the Fathom Results and Conclusion page, with the method, a glossary, and an FAQ alongside it.
What the two projects have in common
Canary and Fathom are separate projects on unrelated subjects, one on AI sector news sentiment, the other on processing complexity in physical and artificial systems. What they share is a method. Both were pre-registered. Both said in advance what would count as failure, and what each possible outcome would mean. Both returned a result that was not the one hoped for, and both reported it plainly rather than dressing it up.
Neither was a commercial product, and neither was intended to be one.
Contact, and access to the record
KnowEntry is maintained by Roger Smolski. Questions, suggestions, and feedback are welcome at roger@knowentry.com.
The complete Fathom record (the pre-registration and its history, every measurement, and the analysis code) is deposited publicly at Zenodo: doi.org/10.5281/zenodo.22160731. The full account of Canary 2.0, including its pre-registered hypothesis and results, is in the Project Conclusion and Final Report. The complete essay, Consciousness and Complexity, appears with a second essay and a full account of Fathom in the book I Am Not Measurable Here (Synaptica Press, 1 January 2027).