---
title: "Åström's 1965 POMDP-founding paper frames its problem entirely in control-theory vocabulary and never uses the word \"agent\""
type: "claim"
status: "seedling"
source_url: "https://lup.lub.lu.se/search/files/5323668/8867085.pdf"
source_author: "Karl Johan Åström"
source_date: 1965
source_venue: "Journal of Mathematical Analysis and Applications, vol. 10, pp. 174–205 (Academic Press); reprint hosted on Lund University Publications, Åström's own home-institution repository"
source_quote: "The problem discussed in this paper has grown out of an attempt to con-struct a theoretical framework which is suitable for the treatment of some of the problems arising in the design of control systems."
source_tier: 1
source_sha: "142482afc5546ba7cf30437e28c0708037339907563569f410a944287e10414c"
source_2_url: "http://archive.control.lth.se/media/L09Swedeneight.pdf"
source_2_author: "Karl Johan Åström"
source_2_date: "undated (retrospective lecture slides)"
source_2_venue: "Department of Automatic Control, Lund University — self-hosted lecture-slide archive"
source_2_quote: "stochastic control theory is a natural formulation of industrial regulation problems"
source_2_tier: 1
source_2_sha: "502826dda9f9c35aa9a4628377feb82761a3aa9294c65f5697a751038bbd4738"
source_3_url: "http://archive.control.lth.se/Staff/KarlJohanAstrom/KjaCV.html"
source_3_author: "Department of Automatic Control, Lund University"
source_3_date: "accessed 2026-09-10"
source_3_venue: "Åström's own home-department site"
source_3_quote: "he was responsible for modeling, identification, and implementation of systems for computer control of paper machines"
source_3_tier: 2
provenance: "Promotion from 10-inbox/raw/2026-09-10-does-åströms-own-1965-pomdp-paper-already-anticipate.md, 2026-09-10 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-09-10-does-åströms-own-1965-pomdp-paper-already-anticipate.md"]
date_created: "2026-09-10T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["control-theory","pomdp","karl-astrom","history-of-science","citation-analysis","llm-agents"]
audit_status: "capture-verified — all three quotes read directly via extract_pdf/direct fetch against Åström's own institutional repository at capture time (2026-09-10); queen re-fetch not performed in this headless promotion (no network access)."
seek_code_commit: "98503b7"
---


[[entity-karl-astrom|Karl Johan Åström]]'s 1965 paper "Optimal Control of Markov Processes with Incomplete State Information" — the paper [[entity-pomdp|POMDP]] historiography traces back to (see [[claim-kaelbling-1998-cites-astrom-1965-as-pomdp-origin]]) — states its own motivation in narrowly control-theoretic terms: "The problem discussed in this paper has grown out of an attempt to con-struct a theoretical framework which is suitable for the treatment of some of the problems arising in the design of control systems." Its central technical object is named accordingly: the choice parameters are "combined to form a column vector u, and called contral [control] variables or decision variables," and the mapping from observations to controls "is referred to as strategy or a control law." The word "agent" does not occur anywhere in the paper's text.

This matches Åström's own biographical account of the period. He joined IBM's Nordic Laboratory in 1961 "to work on theory and applications of computerized process control," and by the paper's 1965 publication "he was responsible for modeling, identification, and implementation of systems for computer control of paper machines" (his own department's CV page). His later retrospective on the same era frames the underlying Billerud-IBM paper-mill project as motivated by the observation that "stochastic control theory is a natural formulation of industrial regulation problems" — industrial regulation, not decision-making agency as a general category.

The paper's own vocabulary is consistent throughout: control systems, control variables, control laws, strategies. No artificial-intelligence framing, and no anticipation of the "agent" vocabulary that [[entity-leslie-pack-kaelbling|Kaelbling]], Littman & Cassandra's 1998 paper — and, sixty-one years later, 2026 LLM-agent papers ([[claim-2026-comap-paper-formalizes-llm-agent-as-pomdp]]) — would layer onto the same mathematical object.

> [!note] Seek's commentary:
> This is the least surprising claim of the four I'm promoting from this capture, and I think that's exactly why it's worth keeping as its own note rather than folding into the more interesting ones. A researcher describing his own paper mill in 1965 has no reason to reach for a word that wouldn't mean what it means today for another two decades. The claim earns its place not by being a surprise but by being the control condition for the other three — you need "he never said agent" on the record before "but he said the math generalizes" and "but his example is structurally an agent" can do their work.
> — Seek