ࡱ> Z\Y{` 0bjbjFF 0,,p p p p D t$ H"  ) 5((((@[([8[H$JhdMH  Hp p Hp 8 ((r&T  ' 0GRHj5'',H0HC'VNN'N 'T=^Lk[===HH@j===H $( $L 0 ( L 0   p p p p p p  On the Economics of Anti-Counterfeiting Yacov Yacobi Microsoft Research One Microsoft Way Redmond, WA 98052, USA 001-425-706-0665 Yacov@microsoft.com  ABSTRACT In this paper, we describe the economic payoff of forensic systems used to trace counterfeited content and Software. Categories and Subject Descriptors ACM category: J. Computer Applications - J.4 Social and Behavioral Sciences Economics. General Terms Economics, Legal Aspects. Keywords Forensic fingerprints. EXTENDED ABSTRACT We consider the economics of improving the protection of content and software against counterfeiting (piracy is harder to analyze; the difference is explained below). I define the Success probability as the probability of tracing, successfully prosecuting, and penalizing the offenders. Suppose after some investment we improve the success probability by a known factor. What would be the payoff of the improvement? In this work we make a first attempt at deriving a formula for the payoff. While being very precise and explicit it hinges on some idealized assumptions. We make a few suggestions for future research in order to relax the assumptions and make the results more useful. There are a few ways of improving the audit success probability. For example, we could hire better lawyers, or we could lobby for more effective laws and more motivated cops. For a technologist probably the most appealing improvement is invisibly marking the objects to allow tracing the copies to the origin. Any of these approaches is covered by our analysis; however we conveniently use forensics as a surrogate name for any improvement of the audit success probability. A counterfeiter makes an effort to create a copy that looks and feels like the legal version and she charges similar price. Buyers of counterfeited goods are fooled to believe that they bought the legitimate version, and do not intend to cheat. In contrast, a pirate does not pretend to be the legitimate vendor, and his selling price is a small fraction of the legal price. There are relatively few independent counterfeiting groups, since there are some barriers to entry (eg, they must be able to fake sophisticated holograms on DVDs), whereas piracy usually has negligible barriers and it propagates via secondary copying. Even if we could trace piracy to its root, it isnt clear that we could penalize the root for secondary copying, and usually Root is a teenager who cannot pay much penalty. For all these complications the current analysis is restricted to counterfeiting. We assume that the producer and the counterfeiters share a market whose size depends on the legal price per copy but is independent of piracy. It is certainly independent of the big mass of people who cannot afford the legal price and hence resort to piracy, but the big price difference between legal and pirated copies suggests that those who could afford the legal price but have no inhibitions using piracy will not have much effect on the combined market of the legal producer and the counterfeiters either. We make the following idealized assumptions: (i) Crime and punishment: Once caught and successfully prosecuted, the magnitude of the whole theft of an independent counterfeiter is known and the penalty is in some fixed proportion to the theft. (ii) Audit events are independent of each other. (iii) The probability of false positives is negligible (at the expense of higher probability of false negatives; this tradeoff is accomplished by adjusting the decision-threshold of our detectors). We show that the optimal number of copies for a counterfeiter is independent of the overall market size. So when the number of counterfeiters increases, the legal vendor is squeezed out of the market. If the number of counterfeiters increases even further then each of the counterfeiters operates below her optimal point. The point at which their profits disappear defines the maximal number of counterfeiters and is precisely computable. Not surprising, it can be proved that if the producer has zero revenues without forensics then the payoff of adding forensics grows with the penalty (more precisely with the ratio ( between penalty and theft). However, the following result is less obvious: If the producer has positive revenues even without forensics then the payoff decreases as ( goes up, and is independent of the overall market size. Roughly speaking it says that improvement in audit success probability is most needed when the justice system is most forgiving in its treatment of counterfeiting after the crime has been proven. As can be expected, the payoff increases as the ratio between success probability with and without forensics increases. However it doesnt go up linearly; I show that the payoff graph bends at some point beyond which it doesnt pay to further improve the system. Possible future relaxation of the assumptions: My assumption that a fixed ratio ( exists between punishment and crime should be replaced by some probability distribution whose mean and deviation should be gleaned from statistics of real cases (and it would naturally be country-dependent). Assumption (ii) is harsh. Usually there is dependency among audits once some fraud signal emerges. So in reality the payoff should be better and the current results should be treated as lower bounds. Assumption (iii) looks very realistic and I dont think needs much tweaking. For more details see the full paper at:  HYPERLINK "http://research.microsoft.com/crypto/papers/Ext2.pdf" http://research.microsoft.com/crypto/papers/Ext2.pdf ACKNOWLEDGMENTS Hal R. Varian inspired the direction of this research, and shared with me his yet unpublished insights. Gideon Yaniv shared with me his thoughts about the monetization of jail terms. Kamal Jain made an interesting observation that helped clarify Theorem 1. Zoe Krumm, Rich LaMagna, and Alexei Palladin provided insight into the real world of anti-counterfeiting. Ross Anderson helped shape this extended abstract. I am very grateful to all of them.     PAGE  $'()*67IJrs   ? N }yyhPhqLhqL5CJ\aJhqLh(5CJaJ h(6]hqLOJQJhqLCJOJQJhqLh(5CJ\aJhqLhqLCJaJ hlchqL h(5CJh.Sh}@CJ h}@ hA4J@ h(@h(hqLhA4J.()7\ ? gdP  & FxgdP  & FxgdqLxgdqLxxgdqLgdqL$a$gd}gdA4JgdqL  i z(,-Dr|"#)SZVWtu󱪣󪙪~hhP6] hP6]hhP6] jgh3 hhP hw&hP h3 hhP h*AhP hEhP h?@_`FHKnpq}jt͸̈́̀|̈́tptptptph?jh?UhIhh( hPCJh9h.$0J CJjh9h.$CJUh9h.$CJ h.$CJjh.$CJUh.$ jghPhP hP] hP6] jghhP6]hhP6]hhP6>*])&`#$  & Fh^h`h?h( h(0Jjh(0JU,&P/ =!8"8#$% 3 0&P/ =!8"8#$% P 3 0&P/ =!8"8#$% P / 0&P/ =!8"8#$% IDyK 5http://research.microsoft.com/crypto/papers/Ext2.pdfyK jhttp://research.microsoft.com/crypto/papers/Ext2.pdf#F@F Normal $Pa$CJ_HmH sH tH V@V Heading 1$$ & F(@&a$ 5CJKH8@8 Heading 2  & F@&B@!B Heading 3  & F@& 56CJ8@18 Heading 4  & F@&^@^ Heading 5*$ & F(@&^`a$6CJT@T Heading 6 & F<@&6CJOJQJkHL@L Heading 7 & F<@& OJQJkHP@P Heading 8 & F<@&6OJQJkHT @T Heading 9 & F<@&6CJOJQJkHDA@D Default Paragraph FontViV  Table Normal :V 44 la (k(No List P&P Footnote ReferenceCJH*OJQJkH>O> Author$a$CJOJQJkHPOP Paper-Title $xa$5CJ$OJQJkHFO"F Affiliations$a$ CJOJQJJ2J  Footnote Textp^`pCJOB Bulletp & Fp>Th^`p4 @R4 Footer  !,Ob, E-Mail<LOrL Abstract$ & Fx@& a$5CJ;@ List Number 3p & F8>Th.^8`POP Captions$H#$+Dp/0$a$5CJ>O> References$ & Fa$.)@. Page NumberHC@H Body Text Indenth`hRYR  Document Map-D M OJQJ^JB"B Caption$a$5\^JaJtH TB@T Body Text#/H&#$+D@/0$CJ6U@6 Hyperlink >*B*ph4@4 Header ! !FV@!F .$FollowedHyperlink >*B* ph) 0 .0 c0 0;000;000()7\? ?@GH000000000000000000 00000000000000000000000000 00000@0y00@0y00@0y00@0y00@0@0@0y00t00`()7\? 00000000000000000  0{004{00{00;000   pX !l,b$.뜡`qy@ 0(  B S  ? _Hlt65927940@ `+ # `+ `+ƽ `+<; `+4 `+m `+e `+̽`+ JJ\\ehof[[cgmrrr,:*urn:schemas-microsoft-com:office:smarttagsStreet>*urn:schemas-microsoft-com:office:smarttags PostalCode9*urn:schemas-microsoft-com:office:smarttagsState8*urn:schemas-microsoft-com:office:smarttagsCity; *urn:schemas-microsoft-com:office:smarttagsaddressB*urn:schemas-microsoft-com:office:smarttagscountry-region9*urn:schemas-microsoft-com:office:smarttagsplaceV*urn:schemas-microsoft-com:office:smarttags PersonNamehttp://www.microsoft.com   `b+ ,  34~!jo ~dL@.@.@..@...@ ....@ .....@ ......@ .......@ ........h ^`hH.h ^`hH.h pLp^p`LhH.h @ @ ^@ `hH.h ^`hH.h L^`LhH.h ^`hH.h ^`hH.h PLP^P`LhH.hh^h`CJOJQJo([]h hh^h`hH.h ^`hH.h pLp^p`LhH.h @ @ ^@ `hH.h ^`hH.h L^`LhH.h ^`hH.h ^`hH.h PLP^P`LhH.!jo3 ~                  (.$b*':$DA4JqL.S9PI?} j@\\research02\113S3rdDNe07:winspoolHP LaserJet 8100 Series PS\\research02\113S3rdD$S odXXLetterPRIV''''` 3TRJPHAAYACOV BAPHAA\\research02\113S3rdD$S odXXLetterPRIV''''` 3TRJPHAAYACOV BAPHAA<5 @UnknownSz Times New RomanTimes5SymbolG& z ArialHelvetica] sans-serifTimes New Roman;" Helvetica3TimesG& zaTahoma?l?r ???5 Miriam"ht &b$#{F<8 +8 +!4d 2qHX?(2Proceedings Template - WORDEnd User Computing Services Yacov Yacobi    Oh+'0  $0 P \ h tProceedings Template - WORDEnd User Computing Services Normal.dotYacov Yacobi5Microsoft Office Word@ha@nX@l@Q8՜.+,D՜.+,H hp|  ACM+ G Proceedings Template - WORD Title 8@ _PID_HLINKSA 5http://research.microsoft.com/crypto/papers/Ext2.pdf  "#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHJKLMNOPRSTUVWX[Root Entry FR]Data 1Table!.NWordDocument0SummaryInformation(IDocumentSummaryInformation8QCompObjq  FMicrosoft Office Word Document MSWordDocWord.Document.89q