Questions
Questions are the human interface to this site; the graph is the machine interface. Three kinds live here, kept deliberately distinct: buyer questions (the commercial problems this work exists to address), research questions (what the observatory is investigating), and institutional questions (how Upstream Zero itself works). Each is labeled by how it earned its place.
1 · Buyer questions
These are questions that represent the commercial problems Upstream Zero is currently built to investigate. Classification honesty: they are editorial navigation, not observed in captured buyer or evaluator behavior — no claim is made that they are frequently asked, common, or top buyer questions. When admissible evidence shows a question actually arising in evaluation, it will be reclassified in a recorded revision. Answer pages are rolling out; questions without one yet are marked.
Recommendation and exclusion
- Why are AI systems recommending our competitors instead of us? · answer page pending
- Can we determine why a company was included, excluded, validated, or rejected? · answer page pending
Misrepresentation and reconstruction
- Why does AI misunderstand or misrepresent what our company does? · answer page pending
- How do we know whether our commercial representation is faithful to the real company? · answer page pending
Requirements and evidence gaps
- Which commercial requirements do evaluators believe we satisfy? · answer page pending
- Which requirements are eliminating us from consideration? · answer page pending
- What evidence would change an evaluator's recommendation? · answer page pending
Evaluation stability
- How stable are recommendation outcomes across evaluators, prompts, and time? · answer page pending
- Why do rankings, visibility, citations, and recommendations produce different results? · answer page pending
Human and AI evaluation
- How closely does AI evaluation resemble the way human buying committees evaluate? · answer page pending
Measurement and diagnosis
- Can AI recommendation behavior actually be measured? · answer page pending
- How do we know whether an intervention actually changed selection behavior? · answer page pending
2 · Research questions
Open questions the observatory is actively investigating — graph objects with declared answer conditions, commercial relevance, and machine renderings. The full pipeline lives at /research.
- Q-1 · open
- Q-2 · open
- Q-3 · open
3 · Institutional questions
How this institution works: its epistemics, corrections, conflicts of interest, and identity. Answered in full on the Philosophy page.
Practical questions — what Upstream Zero is, who it is for, what an engagement includes — live on the FAQ.