AI Glossary / Graphs and symbolic AI / Intermediate
English auction
Definition
An English auction is an ascending-price auction in which participants compete through increasing bids, or remain active as a displayed price rises. The winner is normally the last remaining bidder or the participant with the highest accepted bid, paying the price specified by the auction rules. In AI, it is relevant to automated bidding agents, multiagent coordination, and mechanism design.
What it means
The central feature is an upward-moving price and an opportunity to respond during the auction. Participants receive some information about others' willingness to continue, although the exact information depends on the format. An open-outcry auction and a computer-operated ascending clock are related implementations, not identical procedures.
The rules determine bid increments, withdrawal, ties, reserve prices, eligibility, and the point at which bidding closes. Those details can change strategic behavior. Calling a marketplace an English auction therefore describes its broad structure without fully specifying the decisions an automated bidder should make.
How it works
The organizer defines the item, eligible participants, and payment rules. Bidding starts at an opening price, or a clock increases the price. Participants submit higher bids or indicate that they remain willing to buy. The process ends according to a stated stopping rule, and the winner pays the applicable final price.
In an ideal private-value clock model, each bidder knows its own value independently of information revealed by other bidders. Remaining active until the price reaches that value is a useful benchmark. Common-value uncertainty, budgets, multiple related items, fees, and discrete bidding increments can change the reasoning. A bidding agent must model the actual auction rather than mechanically applying the benchmark.
Three practical examples
- A classroom AI experiment compares agents that leave an ascending clock at their assigned private values with agents that leave prematurely.
- A resource-allocation simulation uses increasing bids to assign one scarce computing slot, with explicitly defined utility and payment rules.
- A marketplace analyst compares open ascending bidding with sealed bids to understand how feedback and bid timing affect the participants' decisions.
Think of it this way
Think of a price thermometer rising until only one participant is willing to stay. This illustrates the ascending clock version and the separation between the winner's value and the price paid. The analogy has limits: real auctions may use individual bids, irregular increments, deadlines, and participation rules rather than a smooth, continuously rising clock.
Worked example
Consider a fictional ideal clock auction for one item. Three agents have private values of 40, 65, and 90 tokens. Assume no fees, no binding reserve, sufficient budgets, and a continuous clock with a specified rule for identifying the last remaining bidder.
The first agent leaves at 40. Two remain until the second leaves at 65. The agent valuing the item at 90 wins and, under this ideal clock convention, pays 65. Its value minus payment is 25 tokens. This is an illustrative payoff, not a measured commercial benefit.
Now change the format to discrete accepted bids. The winning payment can differ because an agent may need to submit the next allowed increment. The example does not establish that every English auction ends at exactly the second-highest value. It shows why the bidding and stopping rules must accompany any calculation.
Illustrative case study
In a fictional university workshop, the Cedar team builds agents for an auction simulator. Its first documentation says only that the highest bidder wins. Test participants then disagree about whether a bid at the closing instant is accepted and whether the winner pays its bid or the runner-up's bid.
The team writes separate specifications for an ascending clock and a sealed second-price auction. It logs clock exits, accepted bids, and payments independently. The teaching improvement is conceptual: learners can explain which rule produced a result instead of assuming that superficially similar auction outcomes imply identical procedures. No real marketplace performance is claimed.
Common misconceptions and mistakes
Do not confuse an English auction with a Dutch auction, where the displayed price typically decreases until a participant accepts. Do not describe an English auction as sealed-bid merely because software submits the bids.
Avoid claiming that truthful behavior, efficiency, or revenue equivalence holds without assumptions. Private values differ from situations where an item's underlying value is uncertain and other bidders possess relevant information. Also distinguish the winner's maximum willingness to pay from its actual payment. Automatic participation does not remove strategic uncertainty or the need for clear rules.
Business use
For a business exploring auction-based allocation, the term helps frame questions about transparency, response time, and agent behavior. A specification should identify the allocation rule, payment rule, information visible during bidding, and exceptional cases.
Evaluation can compare allocations and payments across controlled simulated preferences before any operational use. Include communication failures and ties in the test design. The glossary concept explains a mechanism; it does not establish that auctions are preferable to posted prices or that any particular bidding strategy is commercially suitable.
People also ask
Is an English auction always conducted by a human auctioneer?
No. An electronic platform can implement an ascending process.
Does the winner always pay its full valuation?
No. The payment follows the rules and competing bids; it can be below the winner's value.
Is it the same as a sealed second-price auction?
No. Under certain models their strategic implications are closely related, but their information flow and procedures differ.
Check your understanding
In the ideal clock example, why does the agent with value 90 receive an illustrative surplus of 25 rather than 50?
Answer: The relevant payment is 65, when the second remaining rival leaves. Subtracting 65 from 90 gives 25. The earlier exit at 40 does not end the auction because two agents are still active.
Related terms
- Mechanism design
- Dominant strategy
- Sealed-bid auction
- Multiagent systems
- Utility
- Auction
- private value
- common value
- multiagent system
Learn next
Learn next about sealed second-price auctions. Write the allocation and payment rules for both formats, then repeat the three-agent example with a minimum opening price above 65. State the new rules before predicting whether a sale occurs and what the winner pays.
More in Graphs and symbolic AI
- Elitism - Elitism is an evolutionary-search mechanism that preserves one or more of the best current candidate solutions when forming the next population.
- Empirical loss - Empirical loss is the loss measured on a finite set of observed examples, usually calculated as the average of a per-example loss function.
- End-to-end learning - End-to-end learning trains a system's learnable processing stages together using an objective tied to its final output.
- Entailment - Entailment is a relationship in logic: a set of statements entails another statement when every interpretation that makes the starting statements true also...
- Environment class - An environment class is a collection of possible environments used to specify, study, or evaluate an intelligent agent.
- Epigraph - The epigraph of a function is the set of points on or above its graph.
Sources
Experimental Economics Center, EconPort, English and second-price auction teaching material: https://econport.gsu.edu/content/handbook/auctions/auctionsexperiments/auctionexpenglishandsecond.html Harkavy and Tygar, original auction-protocol paper, auction formats section: https://www.usenix.org/legacy/publications/library/proceedings/ec98/full_papers/harkavy/harkavy_html/node2.html
Previous: Energy in frame Next: Ensemble