Copyright and Technology Conference Observation protocol & impact analysis

2,443 views

Published on

Presentation given by Pauline Blassel, research director of HADOPI, at Copyright and Technology London 2013, October 17, 2013.

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
2,443
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
4
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Copyright and Technology Conference Observation protocol & impact analysis

  1. 1. Copyright and Technology Conference Observation protocol & impact analysis London – October 2013
  2. 2. Hadopi’s missions An independent public authority : Protecting works from copyright infringements (graduated response) ; Encouraging the development of online legal offers ; Regulating the use of technical protection measures ; Observing legal and illegal consumption and use of cultural contents on the Internet.  Raising awareness on copyright and creation issues (ex: educational workshops towards the educative community) 23/10/2013 - 2
  3. 3. The graduated response procedure in a nutshell An awareness-raising mechanism, designed to convert illicit practices into legal consumption Impose penalties only as a last resort 23/10/2013 - 3
  4. 4. The current debates in France regarding our missions and institution The “Lescure report” (on France’s cultural policies to the digital age) The suppression of the one-month suspension of internet access additional penalty A recent attempt to transfer the institution’s missions via an act on public broadcasters A bill to be examined in 2014 23/10/2013 - 4 23/10/2013 - 4
  5. 5. Observation protocol
  6. 6. Observing legal and illegal uses on the Internet Lack of data Technical & legal constraints Definition of a work schedule (contents, circulation, user) Gathering : Qualitative methodologies Quantitative methodologies Computer science research 23/10/2013 - 6
  7. 7. Application case : YouTube status Qualitative methodologies « Digital Natives » Various perception of illegality depending on the practice « Well you know, I mean I don’t know, I think it is illegal sometimes on YouTube as well. I can listen to unlimited music for free » (22-24) « When you get full recent movies, you can wonder whether it is actually legal » (19-21) 23/10/2013 - 7
  8. 8. Application case : YouTube status Computer science research « Streaming platforms contents taxonomy » Selected on the basis of their popularity in France (YouTube and Dailymotion) Representative sample (3000 videos) obtained by random walk Disclaimers Suggested and public videos No localisation Manual categorization of the content 23/10/2013 - 8
  9. 9. Application case : YouTube status Computer science research « Streaming platforms contents taxonomy » Music videos are by far the most popular. 25% of them are « official ». Next in popularity come the movies (full or organized in sequences) (2,6%) 23/10/2013 - 9
  10. 10. Application case : YouTube status Quantitative methodologies Consumption diaries On average, music consumers have listened to 17.4 titles on streaming websites and downloaded 6.4 titles. Weekly individual average consumption for each cultural good, in France (nombre de biens culturels sur une semaine) Consommateurs** (53761) 17,1 (morceaux de musique/vidéo clips) Streaming 2,3 (3171) 2,8 (6134) 2,3 (1242) - **Consommateurs de biens culturels de la catégorie concernée au moins une fois au cours de la semaine de référence 23/10/2013 - 10
  11. 11. Application case : YouTube status Quantitative methodologies Consumption diaries YouTube is widely used among respondents to access digital cultural contents : Streaming Download Music 54% 25% Movies 18% 3% TV shows 7% 2% Books 2% 1% 54% of the respondents who stream music are using music in streaming. to access 23/10/2013 - 11
  12. 12. Impact analysis
  13. 13. Bias Representativeness and significance  The use of samples  For the results to be relevant, the sample used must satisfy two requisites: To be representative of the “system” studied To be “large enough” (n) to be significant ( 2 × 𝑒𝑟𝑓 −1 (1 − 𝑒))2 × 𝑝(1 − 𝑝) 𝑛= 𝑒2 23/10/2013 - 13
  14. 14. Bias Correlation and causality Correlation between X and Y is computed as follows: 𝐶𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛(𝑋, 𝑌) = 𝑥− 𝑥 𝑥− 𝑥 2 𝑦− 𝑦 𝑦− 𝑦 2 Correlation means that X and Y evolve: in the same way (if the correlation is positive) or in the inverse way (if the correlation is negative) in different ways (if correlation is near to zero) Possibilities X causes Y Y causes X Y causes X and X causes Y A third factor causes X and Y Coincidence 23/10/2013 - 14
  15. 15. Correlation and causality • Those two variables are clearly correlated • But surely no cause to effect link between them 4 P2P apps (non BT) Audience UV 4 P2P apps (non BT) Australian coal metric ton price Correlation = 0.763 Source : comScore MMX, France, sample of French internet users aged 6+, oct. 2012 – sept. 2013. P2P apps category as custom-defined by Hadopi (BearShare (App), Shareaza (App), eMule (App), Ares Galaxy - P2P (App)). Source : http://www.indexmundi.com 23/10/2013 - 15
  16. 16. Misinterpretation and extrapolation Attributing more sense to a measure than it actually has. Can’t be mathematically modeled. Example on one Torrent directory portal : no relation proven between global piracy and sites UV audience Pages Viewed (MM) Unique Visitors (M) 3 600 240 3 400 220 3 200 200 3 000 180 2 800 160 2 600 140 2 400 120 2 200 100 Source : comScore MMX, France, sample of French internet users aged 6+, oct. 2012 – sept. 2013 One Torrent Directory portal. 23/10/2013 - 16 23/10/2013 - 16
  17. 17. Bias Misinterpretation and extrapolation 26 DDL platforms Total pages viewed With or without Mega « possible curiosity effect » Summer period « Possible circumstances effect » Source : comScore MMX, France, sample of French internet users aged 6+, june 2012 – august. 2013 DDL plateforms category as custom-defined by Hadopi, including : 1FICHIER.COM, UPLOADED.NET, TURBOBIT.NET, PUTLOCKER.COM,RAPIDGATOR.NET, UPTOBOX.COM, MEDIAFIRE.COM, MEGA.CO.NZ, UPLOADING.COM, HOTFILE.COM, MEGASHARES.COM, etc. 23/10/2013 - 17
  18. 18. Impact analysis - Big picture Decline in P2P use in France after Hadopi implementation according to IFPI’s data 7 000 000 users 6 500 6 000 5 500 Deeper observations are needed to analyze this recent trend. 5 000 4 500 4 000 3 500 According to these IFPI’s data, with no comments on the methodology – considering we are talking about gathered data – we observe 3 trends, the less significative of which needs a deeper analysis : Strong decrease the first year Inflexion in September 2011 Remains limited Source : IFPI (Nielsen Data) 23/10/2013 - 18
  19. 19. Impact analysis – Focus 2012-2013 Torrent & other P2P Apps (Unduplicated UV) 8 200 000 8 000 7 800 7 600 4 BT apps 7 400 1 UV = the app starts running at least once (including passive uses) for an individual in a month. 7 200 P2P apps audience since July 2012 4 600 4 400 000 4 200 4 000 3 800 3 600 4 P2P apps 3 400 3 200 Source : comScore MMX, France, sample of French internet users aged 6+, juil. 2012 – sept. 2013 BT apps (uTorrent (App), Azureus (App), BitComet (App), BitTorrent(App)) and P2P apps (BearShare (App), Shareaza (App), eMule (App), Ares Galaxy P2P (App)) categories as custom-defined by Hadopi 23/10/2013 - 19
  20. 20. Impact analysis – Focus 2012-2013 Torrent directory portal (UV) (1/3) 5 600 5 400 000 5 200 5 000 What could this trend observed specifically on torrent directory portals mean ? 4 800 4 600 4 400 More details are needed for a straight conclusion Unduplicated UV 8 Torrent directory portals 4 200 It shows a raise in the number of unique visitors on torrent directory portals But doesn’t inform on their uses when visiting those portals (quitting or downloading) Considering the biases listed, it does need more details to conclude (Case by case basis and pages viewed). Source : comScore MMX, France, sample of French Internet users aged 6+, juil. 2012 – sept. 2013 Torrent directory portal category as custom-defined by Hadopi (T411.me, CPasBien.com, TorrentFrancais.com, SmarTorrent.com, IsoHunt.com, OmgTorrent.Com, TorrentReactor.net, Mininova.org) 23/10/2013 - 20
  21. 21. Impact analysis – Focus 2012-2013 Torrent directory portal (ADV) (2/3) 600 500 Average Daily Visitors Case by case basis 000 400 300 200 100 0 Actually : On a case by case basis (on the 8 major torrent directory portals) the number of average daily visitors remains pretty stable Source : comScore MMX, France, sample of French internet users aged 6+, june 2012– sept. 2013 Torrent directory portal category as custom-defined by Hadopi 23/10/2013 - 21
  22. 22. Impact analysis – Focus 2012-2013 Torrent directory portal - Pages viewed (3/3) 400 350 000 300 250 200 150 100 50 Pages viewed 8 Torrent directory portals Global piracy may be different from UV audience 0 Moreover, the number of pages viewed remains stable as well Pages viewed could be the most relevant indicator about a volume of piracy. A visitor is not a downloader, and one visit can find various explanations. It highlights the need of relevant data to analyze piracy and the impact of public policies Source : comScore MMX, France, sample of French internet users aged 6+, june 2012– sept. 2013 Torrent directory portal category as custom23/10/2013 - 22 defined by Hadopi (T411.me, CPasBien.com, TorrentFrancais.com, SmarTorrent.com, IsoHunt.com, OmgTorrent.Com, TorrentReactor.net, Mininova.org) 23/10/2013 - 22
  23. 23. Stability is not enough Hadopi’s current projects Meanwhile, in addition to the observation task, the institution as a whole is working on « Outside the box » possibilities : Strengthening the fight against commercial counterfeiting ?  illegal streaming and direct download websites (& Bit Torrent sharing protocol)  websites themselves  technical and financial intermediaries Assessing the ability to model a system of proportional remuneration of sharing ? Websites generating incomes with the content shared (not end consumers)  Different from a « voluntary collective licensing » or a « global license » (a proportional amount versus a flat-fee)  In partnership with a French public research body fully dedicated to computational sciences (INRIA)  Need to address technical, financial and legal issues.  23/10/2013 - 23

×