Online Auction Fraud Dr Jarrod Trevathan Discipline of IT, School of Business.

Online Auction Fraud

Dr Jarrod Trevathan

Discipline of IT,

School of Business

Overview

• The online auction format

• Types of online auction fraud and its effects on auction users

• What are online auctioneers doing about it

• Where the law sits on the topic

• Our research into fraud detection/prevention mechanisms

• Future work and problems to overcome

Online Auctions

• Since the mid 90s, online auctions have continued to increase in popularity– Ability to reach a world wide audience– Can participate from home/work– Wide availability of items and collectables

Online Auction Fraud

• One of the fastest growing areas for Internet fraud complaints

• Internet Crime Complaint Centre:– Second most reported offence– Accounts for 25.5% of a total of 275,284

complaints– Resulted in a financial loss of $265 million

Online Auction Fraud

• Easy to commit:– Participants are largely anonymous– Lack of accountability– Auctioneers don’t actively police auctions– Law is largely undefined and impotent

Types of Auctions

The Online Auction Format

• Auctions terminate at a given time

• Bidders can submit bids at any time prior to the closing time

The Online Auction Format

• Proxy bidding systems:

– A bidder lists the maximum price s/he is willing to pay

– If outbid, the system will submit minimal bids up to the maximum to stay ahead

– A bidder doesn’t have to monitor the auction

• Auction fraud is geared towards providing a specific participant with an unfair advantage

Online Auction Participants

Seller Auctioneer Bidders

Observers

Bid Shielding

• Collusion among bidders to keep the price low• Disadvantages the seller and the auctioneer

Bid Shielding

• Proxy bids affect the process

• Pooling, bid rigging

Shill Bidding

• Using fake bids to inflate the price

• Bidders are referred to as shills

• If the shill wins, the item is resold

• Shilling is strictly forbidden

• It is a prosecutable offence

– March 2001, U.S. Federal Grand Jury– Art auctions– 40 different aliases

Shill Mindset

• Goal: Inflate the price for the seller

• Increase bid, increase revenue

• Increase price, increase chance of failure

• Shill has an infinite budget, and bids to lose

• Bidder has a finite budget, and bids to win

Shill Strategies

We have identified the following strategies:

1. Bids exclusively in auctions only held by one particular seller

2. High bid frequency

3. Few or no winnings

4. Bids quickly after a rival bid

5. Only bids the minimal amount required

6. Does not bid near the end of an auction

Aggressive Shill Bidding

• Shill aggressively bids throughout the auction

Benign Shill Bidding

• Shill makes a one-off bid at or near the beginning of the auction

Shill Detection

• Confusion over shill behaviour

• Commercial auctioneers do not make their detection methods public– State they have mechanisms but do not

reveal what they are or how they work– ACCC viewpoint– We have been contacted by eBay’s layers in

Nashville Tennessee

Shill Detection

• Wang et al, 2002 suggest listing fees

• Shah et al 2003 use basic data mining techniques

• Xu uses a concurrency model

• We propose the shill score (Trevathan and Read, 2005)

• Bidders are given a score between 1 and 10• The higher the score, the more suspicious• Bidders can decide if they want to participate

The Shill Score

• Bidders are assigned a score between 0 and 10

α - number of auctions participated in

β - number of bids per auction

γ - wins to auction participation rate

δ - speed of bidding

ε - size of bid increment

ζ - normalised time bidder commences bidding at

αθ1 + βθ2 + γθ3 + δθ4 + εθ5 + ζθ6

Shill Score = ————————————— θ1 + θ2 + θ3 + θ4 + θ5 + θ6

The Shill Score

Bid Sniping

• Bidder bids in the closing seconds

• Rivals are denied time to react

Bid Sniping

• Permitted on eBay, but discouraged

– Seller loses as the price is not as high as it

could have been

– The auctioneer may lose commission

– Bidders are upset

• Opposite of shilling

– used as a defence against shilling

Bid Sniping

• Commercial sniping agents

• Defences against bid sniping

– Out snipe the sniper• Results in a rally between snipers

• The person with the fastest connect usually wins

– Random time-out extensions• The auction is kept alive for each new bid received

Bid Siphoning

• Outsider observes an auction• Offers bidders an identical item for less

– Avoids the costs of auctioning

• This is undesirable as:– Seller loses revenue and customers– The auctioneer loses auction fees– Can be used in conjunction with shilling or

other types of scams

Bid Siphoning

• eBay recommends:

– Bidders should report siphoners– Communication (i.e., email) can only be

between registered users– Under no circumstance deal with the siphoner

as trades outside the normal mechanism will void insurance and protection

Bid Siphoning

• eBay’s bidder masking policy

• Doesn’t deliver the item or misrepresents it to be of higher value

Misrepresented or Non-Existent Items

Misrepresented or Non-Existent Items

• eBay recommends:

– Inspect photos• The seller might use fake photos

– Ask questions• The seller might not respond• The seller can lie• The bidder might want to remain anonymous

Misrepresented or Non-Existent Items

• eBay recommends:

– Seller feedback ratings• Existing reputation systems are dubious

– Dispute resolution procedures• May not fully compensate victim

– Insurance• Recourse can be lengthy and expensive• Only taken after fraud has occurred

Misrepresented or Non-Existent Items

• Escrow Fraud

– Check for poor grammar

– Site usually copied from a legitimate site

– Give-aways in the Terms page

– Evidence left of previous incarnations

– Sellers don’t usually press for escrow

Buyers Do

Misrepresented or Non-Existent Items

• Feedback rating system

Misrepresented or Non-Existent Items

• Feedback rating system– Many researchers have criticised the formula used and the

effectiveness of this system

– Over 95% of sellers have a feedback rating of at least 97%

– Feedback Farming

• Get 100 accounts and hold 100 auctions for cheap items – instant

positive feedback for minimal cost!

– Pack Wolf Feedback Attack

• Automated bots provide a stream of negative feedback to destroy a

seller’s credibility

Conclusions

• The online auction format blurs the lines between what the auction

type is, and what strategy to use

• This has introduced the possibility for numerous types of fraud

• New types of fraud continue to emerge

• Technical and legal measures are vague and ineffective

• Casts doubt over online auctions as a trading mechanism

Caveat Emptor / Caveat Vendor

Our Book

Questions