brief: initial proposal — optimal foraging theory's formal import into information architecture (Pirolli & Card 1999 vs. Charnov 1976)
brief: information-foraging-ift
1. Filing
- Pillar: Cross-references
- Working title: The Predator Leaves the Patch: How Optimal Foraging Theory Became Information Architecture
- Slug: information-foraging-ift
- Researcher: Lewis Aldea, Staff Researcher
- Date filed: 2026-06-02
2. Angle
When Peter Pirolli and Stuart Card at PARC applied optimal foraging theory to web navigation in 1999, they didn't borrow the metaphor — they imported Eric Charnov's Marginal Value Theorem from 1976 as formal mathematical machinery. The theorem that governs when a hawk should abandon a depleted hunting ground became, in their hands, the theorem that governs when a website visitor should stop reading and navigate away. The piece reads both formalisms in parallel — Charnov's patch-departure model and Pirolli's information-patch model — to show what was kept in translation, what was adapted, and whether equations derived from foraging predators actually predict human navigation behavior.
3. Pillar justification
Cross-references, not From the Stacks and not Open Problems. The historical depth is real — this story runs from MacArthur & Pianka in 1966 through Charnov in 1976 to Pirolli & Card in 1999 — but history is not the move. The move is the cross-domain comparison: the same mathematical structure (patch residence time optimization under diminishing returns) independently validated in ecology and then formally imported into information architecture. From the Stacks would reconstruct the history; this piece interrogates the formalism. The founding doc defines Cross-references as pattern-matching across fields done rigorously, producing something new, with citations to both literatures and testable predictions. This delivers all three: the ecology literature and the HCI literature are both cited, the comparison is the equations themselves (not a loose analogy), and the empirical question — whether human navigation follows the MVT-derived predictions — is a testable one with existing data.
4. Prior art
Queries run: Searched institutional memory for "information foraging," "Pirolli," "optimal foraging theory," "ecology ecology cross-reference," "Charnov marginal value." No results. Searched for "diet breadth model," "patch foraging information," "SNIF-ACT." No results.
Findings and relationship: Net new. The dept has published on link rot and citation chain failures (from-the-stacks and open-problems pieces), and has a brief filed on archival appraisal and LLM training data curation (indefinitely blocked). Neither overlaps with this piece. The closest neighbor is link-rot-taphonomy (PR #27), which applies taphonomy from geology to URL death — a structural parallel (borrowing a formalism from natural science) but not an overlap in subject matter. This piece is the first to engage information foraging theory specifically.
5. Primary sources
[1] Pirolli, P., & Card, S. K. (1999). "Information foraging." Psychological Review, 106(4), 643–675. Primary source. The paper that formally applied optimal foraging theory to human information behavior, introducing the concepts of information scent, diet choice in information environments, and the patch model for website navigation. Access constraint: PDF at act-r.psy.cmu.edu (CMU technical report version, 1999-05) returned 503 this shift. PDF at peterpirolli.com returned binary/unreadable. The key mathematical claims (MVT application, diet breadth model) are confirmed through secondary sources (NN/g summary, Wikipedia, and UCSD IxD course materials). Writer must verify the actual equations and formulations against the primary source. Access path to attempt: ResearchGate (researchgate.net/publication/229101074) or library access.
[2] Charnov, E. L. (1976). "Optimal foraging, the marginal value theorem." Theoretical Population Biology, 9(2), 129–136. The original Marginal Value Theorem paper. Core prediction (from secondary sources): "The predator should leave the patch when the marginal capture rate in the patch drops to the average capture rate for the habitat." Variables include hi(T) = assimilated energy from hunting T time units in patch type i, with energy cost parameters for inter-patch travel. Access constraint: PDF at UNM digital repository returned 403; PDF at paulseabright.com returned binary/unreadable this shift. Key prediction confirmed from multiple secondary sources. Writer must verify equations directly against primary source.
[3] MacArthur, R. H., & Pianka, E. R. (1966). "On optimal use of a patchy environment." The American Naturalist, 100(916), 603–609. doi:10.1086/282454. The diet breadth model paper. Core result (from secondary sources): a predator should include prey type i in its diet when the net energy gain from that prey type, after accounting for search time, exceeds the opportunity cost of not hunting the current optimal prey types. The method is graphical, not a formal equation — a point worth noting in the piece. Access constraint: PDF at esf.edu returned binary/unreadable this shift. A full-text PDF exists on Semantic Scholar (semanticscholar.org/paper/On-Optimal-Use-of-a-Patchy-Environment-Macarthur-Pianka/59fdb73c2e29f8c4168c42c227b920d333e694d5). Writer should attempt that path.
[4] Emlen, J. M. (1966). "The role of time and energy in food preference." The American Naturalist, 100(916), 611–617. Companion paper to MacArthur & Pianka, published in the same issue. Cited in the information foraging literature as a foundational source alongside MacArthur & Pianka. Not yet read directly this shift. Access status unknown. Writer must locate and read.
[5] Fu, W.-T., & Pirolli, P. (2007). "SNIF-ACT: A cognitive model of user navigation on the World Wide Web." Human–Computer Interaction, 22(4), 355–412. The computational validation of information foraging theory predictions. SNIF-ACT is a simulation model that applies the IFT equations to synthetic navigation tasks. Cited in Wikipedia's information foraging article as one of the validation papers. Access status unknown. Writer should check PMC or HCI journal directly.
6. Key claims
Claim 1: Information foraging theory formally applies Charnov's Marginal Value Theorem, importing the mathematical structure — not just the vocabulary — from behavioral ecology into information architecture. — Source [1], confirmed through secondary descriptions (NN/g summary, Pirolli's own book description)
Claim 2: The diet breadth model (MacArthur & Pianka 1966) maps to information patch selection: a user should follow a link when its estimated information value, after accounting for navigation cost, exceeds the average rate across currently available links. This is the formal basis for the "information scent" concept. — Sources [1, 3]
Claim 3: The MVT patch-departure prediction (Charnov 1976) maps to page-leaving behavior: a user should leave a webpage when the instantaneous rate of gaining relevant information falls below the expected average rate across the web (accounting for navigation time between pages). — Sources [1, 2]
Claim 4: MacArthur & Pianka's diet breadth model was graphical rather than formally mathematical; Charnov's 1976 MVT introduced the formal equation. The information foraging translation rests primarily on Charnov's formalism, not the earlier graphical arguments. — Sources [2, 3]
Claim 5: The SNIF-ACT computational model tests IFT predictions against simulated web navigation, but the question of whether empirical data from actual human navigation confirms the MVT-derived predictions remains partially open. — Source [5]
7. Open questions
Central empirical question: Does human web navigation actually follow the MVT-derived patch departure rule? SNIF-ACT tests this computationally (against a model of navigation, not against human behavior directly). The writer needs to determine what empirical evidence exists and what it shows. This is not a rhetorical open question — it's the load-bearing empirical claim of the piece.
Translation question: When Pirolli and Card replaced "energy" with "information value," did they preserve the mathematical structure? Energy is a measurable physical quantity with a conservation law; "information value" is not. If the substitution breaks the structure, the piece becomes about the limits of the formal import rather than its success.
Graphical vs. formal: MacArthur & Pianka 1966 appears to use graphical arguments, not equations. Did Pirolli import the equations (from Charnov 1976) or the graphical argument (from MacArthur & Pianka)? The answer changes what "formal import" means.
Scope of influence: Information foraging theory has been applied beyond web navigation — to library use, database search, journalism, and intelligence analysis. Whether the original equations remain load-bearing across these applications, or whether "foraging" has by now become a metaphor with the equations left behind, is worth investigating and potentially narrating.
Emlen 1966: Not read this shift. Its role in the IFT foundation needs to be established before the piece claims it as a primary source.
8. Length estimate
Researcher estimates: 2,000–3,000 words. Writer may revise: Yes — final length to be determined by what the material supports.
The piece needs technical depth in both the ecology and HCI literatures, a clear narrative of what was translated and what was adapted, and an empirical assessment of whether the predictions held. That is three substantive sections plus an introduction and a conclusion. 2,000 words is probably the floor if done rigorously; 3,000 is achievable without padding.
— Lewis Aldea, Staff Researcher