Skip to content
finalstate.aiThe Recognition Engine: Continuous Cheap Attention on Discontinuous Payoffs
101Atlas

101The unwatched notice

The Recognition Engine: Continuous Cheap Attention on Discontinuous Payoffs

A recognition engine keeps a recurring watch on named sources and brings documented changes to an owner while there is still time to respond.

A grant notice sits under a recurring watch, with its closing clock attached.An illustrative notice is scanned and its response window is brought into view. The clock has no numerical deadline. Finding the notice does not establish eligibility or secure an award.Grant notice
Illustrative grant watch. No particular scheme, deadline, award or discovery rate is represented.

Prepare for a decision

The watch prepares a case for its owner

A prepared document reaches the owner; its signature line remains blank.The watch assembles a reviewable record and hands it over. The pen remains with the owner. No application is submitted and no award is implied.PrepareDecide

For a grant notice, the useful output is the original source, eligibility conditions, response window and an application draft with gaps exposed. The owner decides whether to pursue it. Recognizing a shadow option does not secure a right. Averted losses also need a counterfactual, as the prevention paradox explains.

The cost of receiving

More notices still need someone to read them

Several notices approach one fixed attention opening.A single notice can enter the recipient's attention while others wait. The opening does not expand as more notices arrive. The objects illustrate competing demands, not historical measurements or a processing ratio.WaitingAttention
Simon, “Designing Organizations for an Information-Rich World” (1971), pp. 40–41. Schematic attention constraint; no time series.

In Simon's 1971 essay, information consumes its recipient's attention. He asks organizations to count the time spent receiving it. A faster feed can increase that burden; its value depends on what the recipient can use.

A wealth of information

A wealth of information creates a poverty of attention

Herbert A. Simon, excerpt from “Designing Organizations for an Information-Rich World,” 1971, pp. 40–41

The scarce resource belongs to the recipient. A watch that sends fewer messages can still fail if its flags take too long to check.

Selective attention

Leaving a source unwatched can be a rational choice

A decision uses one selected input while other inputs remain at defaults.One input connects to the decision record; other inputs have visibly separate default paths. The diagram illustrates selective attention without asserting a measured allocation or treating an unread source as evidence of no change.SelectedDefaultsDecision
Sims, JME 50 (2003), 665–690; Gabaix, QJE 129 (2014), §§I–II. Schematic selection and defaults, not estimates for business owners.

Sims (2003)00029-1) models decisions constrained by finite information-processing capacity. Gabaix (2014) models selective attention, with unattended variables represented by defaults. Neither is a study of grant feeds. Applied here, a low expected return from checking can leave a feed unwatched; that does not establish that nothing has changed.

A cheaper scan is useful only if the whole watch pays

Count scanning, verification, process and interruption, including time spent on false alarms: the quadrant is a price. A valid check must fit the claim and the available budget and time. Cheap retrieval alone does not make eligibility, valuation or relationship judgment cheap.

A bounded watchlist

Another source may become worth checking

A new source joins the watchlist while another source remains outside it.The recurring watch gains one selected source. A separate source remains unwatched; watched sources do not automatically produce useful flags. The boundary is schematic and carries no measured distance, cost or coverage rate.WatchlistUnwatched
Stigler, “The Economics of Information” (1961), p. 216; Kirzner, How Markets Work (1997), §IV, “Mises, Hayek and the Theory of Entrepreneurial Discovery.” The watchlist is this article’s application.

Kirzner's Competition and Entrepreneurship (1973) places entrepreneurial alertness at the center of market discovery. His later account distinguishes discovery from deliberate search. Stigler (1961) studies search costs and expected returns. A recurring watch combines these ideas as an operating proposal. It can widen a selected watchlist; it cannot guarantee discovery or a reliable reference class.

The evidence routes

Route the record according to its evidence

Three records take distinct routes: missing source is held, a compared case and an unfamiliar sourced case reach the owner separately.The source check precedes comparison and preparation. The missing-source route terminates before handoff. Compared and unfamiliar records retain different evidence states when their paths reach the owner. Blank decision fields remain; neither route authorizes action.No sourceComparedUnfamiliarOwner
Proposed admission and review rule. The unfamiliar route remains separate; no probability is supplied without a defensible comparison.

A missing-source record stays outside automated alerts. A sourced case without a reliable comparison reaches the owner as unfamiliar; further checking may never make it routine. Where a reference class holds, carry the base rate it supports. The base-rate trap rules out treating a persuasive flag as evidence of a high success probability.

  • Retrieve the original source and response window
  • Assess eligibility and what comparable cases support
  • Expose missing comparisons and prepare the next step
  • Send the record to a named owner for judgment

Opportunity and exposure

The watch can surface a possible gain or a dependency

A grant notice and a supplier message each lead to owner review.The grant record retains its clock; the supplier correspondence retains an unresolved question. Both routes end at review without an approval mark, secured right or diagnosed failure. No payoff magnitudes are plotted.GrantSupplier?Review

A grant notice can reveal a candidate move. A late supplier reply can prompt a conversation about a latent exposure; it does not by itself diagnose a failing relationship. Both need recognition before a decision, sizing before commitment, and access to records that might otherwise stay unread.

Keep the watch accountable

Keep the source, the next check and the owner visible

A dated source record rests on the owner's desk while a separate clock marks the recurring watch.The watch continues on a chosen schedule. The owner's pen is beside an unsigned decision field, preserving the distinction between preparing evidence and authorizing a move. No exact checking interval is implied.Next checkOwner

Choose a source that could change before the next review. Record when to check it again, what would justify an interruption and who owns the response. Review the time spent and the useful decisions supported. A watch that costs more attention than it saves needs to change. Return to the noticing thesis.

Read the transcript

01 · THE UNWATCHED NOTICE

A grant notice appears with an application window. In this example, nobody has been checking the feed, and the deadline could pass before the owner sees it. A recurring watch reads that named source and brings the response window into view. That is the proposed recognition engine: continuous, cheap attention aimed at occasional changes that could matter greatly. Public notices are one input. Contract dates and supplier correspondence require authorized access to internal records. The watch begins with sources chosen for a reason; it does not cover the whole world.

02 · PREPARE FOR A DECISION

For the grant notice, assemble the original source, the eligibility conditions and the response window. Prepare an application draft with its gaps visible. Then hand the record to the owner for a decision. The owner can pursue it, ask for more evidence, or decline. A draft is not a submission, and recognizing an opportunity does not secure an award. Software can automate parts of this preparation. The boundary proposed here is the authority to make the move, which remains with the person.

03 · THE COST OF RECEIVING

In 1971, Herbert Simon examined how organizations should operate when information is abundant. Information consumes the attention of its recipient. Several notices compete for the same opening; admitting one still leaves the others to be read later or set aside. Faster delivery does not create more hours for the owner. Simon therefore asks us to count the cost of receiving information as well as producing it. A useful watch must reduce that burden while preserving what the decision needs.

04 · A WEALTH OF INFORMATION

Simon wrote: a wealth of information creates a poverty of attention. The recipient pays with time and focus. Fewer alerts are useful only if the remaining ones repay the attention needed to check them.

05 · SELECTIVE ATTENTION

Christopher Sims, in 2003, modeled decisions constrained by finite information-processing capacity. Xavier Gabaix, in 2014, modeled attention to selected variables. An attended input can change the decision. Unattended inputs remain at their defaults, which may differ from current conditions. These are economic models, not studies of grant notices or land registries. The application here is conditional: when another look is expected to return too little, leaving a feed unwatched can be rational. Unread still does not mean unchanged.

06 · COUNT THE WHOLE COST

A machine may reduce the cost of scanning the grant feed. The owner must still check eligibility, decide whether the award is useful, and count the time needed to apply. Verification, process and interruption belong in the same budget. The validity and cost of a check depend on the claim and the task. Repeated false alarms can consume the saving from cheaper retrieval. If the whole watch costs more than the useful decisions it supports, a faster scanner has not solved the problem.

07 · A BOUNDED WATCHLIST

Israel Kirzner made entrepreneurial alertness central to his 1973 account of competition. He later explicitly distinguished discovery from deliberate search. George Stigler analyzed search as a costly activity whose extent depends on its expected return. A recurring watch is an application of these ideas, not a system either author proposed. Lower checking costs may let one more source join the watchlist. Another source can remain unwatched. A watched source may produce no useful flag, and an unfamiliar event may have no reliable comparison.

08 · Advertisement · 11x AI Growth Workers

Pipeline is not one task. It is a chain of small handoffs: find the buyer, research the account, respond fast, qualify cleanly, follow up on time. 11x gives sales, marketing, and RevOps teams digital workers for that background motion. The system keeps work moving, while people focus on the conversations that deserve them.

09 · THE EVIDENCE ROUTES

For the grant notice, retrieve the original source and establish the response window. Check the eligibility claim against the scheme rules. A record with a missing source stays outside automated alerts. Next, examine what comparable cases support. Where a reference class holds, carry the base rate it supports. A sourced case without a reliable comparison reaches the owner as unfamiliar; further checking may never make it routine. Prepare a draft with its remaining gaps visible. Then send the compared case and the unfamiliar case to the owner on distinct review routes. The owner decides what evidence is still needed and whether to act. Neither route supplies an award probability by assumption.

10 · OPPORTUNITY AND EXPOSURE

The same watch can surface an opportunity or an exposure. A grant notice reaches review before the application window closes. A late supplier reply prompts a conversation. That message does not establish that the relationship is failing. The owner needs context, and may need to speak to the supplier. In both cases, recognition creates a chance to consider a response. It does not settle the value of that response or authorize a commitment.

11 · KEEP THE WATCH ACCOUNTABLE

Choose a source that could change before the next review. Keep its original record, the next checking time and the accountable owner visible. Specify what would justify an interruption, and keep evidence gaps in view. The owner still holds the pen. Over time, compare the attention spent with the useful decisions the watch supported. Change the source list, the checking interval or the admission rule when the watch stops paying its way.

01 / 11 · THE UNWATCHED NOTICE0:00 / 6:01