Slide 1: Creative Commons Statistics from the CC-Monitor Project
Giorgos Cheliotis School of Information Systems Singapore Management University giorgos@smu.edu.sg
Based on a presentation at the iCommons Summit, Dubrovnik, June 14-17, 2007
Slide 2: License (1/2)
This presentation* is licensed under a Creative Commons license: http://creativecommons.org/licenses/by/3.0/
* with the exception of the slide layout and the SMU logo which are property of SMU
This material is released early due to high demand and for the benefit of the Creative Commons community – researchers and academics interested in the details of the work are advised to contact giorgos@smu.edu.sg, as the related research is ongoing and currently in the process of being published. See next page for license details…
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
2
Slide 3: License (2/2)
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
3
Slide 4: Motivation for our study of CC
Before CC most content authors were faced with a binary decision problem: reserve all rights (default copyright protection) or give it all up (public domain) With CC for the first time we can observe large numbers of users making conscious licensing decisions for their content! First-level questions • How many authors use CC? • Who are they? • Which licenses do they prefer? • What is the impact of their choice? • How do jurisdictions compare? • Which factors influence this valuation? • What are suitable business models for CC content? The really important questions • How strong is CC adoption? • How do users value different rights?
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
4
Slide 5: Estimates of CC license popularity
• Some data has been made available online by Mike Linksvayer and Christian Ahlert (Openbusiness), in a paper by Zachary Katz, and in a user survey documented in the PhD dissertation of Minjeong Kim • Most data collection efforts based on Yahoo and Google search results • Some observations made in the past:
– – – – – Non-BY licenses barely used (and therefore dropped) Total of millions of CC-licensed items (various estimates) NC licenses more popular SA and ND also popular attributes Media type may play a role in licensing (music more liberal)
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
June 14, 2007
5
Slide 6: Data collection process (simplified)
Backlinks (Yahoo Site Explorer) Yahoo
YBL
Analysis
Usage Data
Data Querying Scripts
Yahoo CCSearch
YCC
Google
• With YBL and GBL we count the number of HTML pages linking to each CC-Deed page • With YCC we use Yahoo’s search for CC metadata
Backlinks (Google Advanced Search)
GBL
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
6
Slide 7: Total volume and license mix
YBL (Total: 37.1m)
16,000,000 14,000,000 12,000,000 10,000,000 8,000,000 6,000,000 4,000,000 2,000,000 BY BY-SA BY-ND BY-NC BY-NC-SA BY-NC-ND
GBL (Total: 1.2m)
13,355,702 (36%)
600,000 500,000 400,000 300,000 200,000 100,000 BY BY-SA BY-ND BY-NC BY-NC-SA BY-NC-ND
478,377 (38%)
4,368,793 (12%)
6,571,412 (18%) 3,749,091 (10%) 987,502 (3%)
8,038,317 (22%)
144,059 (11%)
222,810 (18%) 98,369 (8%)
286,259 (23%)
38,427 (3%)
YCC (Total: 14.4m)
7,000,000 6,000,000 5,000,000 4,000,000 3,000,000 2,000,000 1,000,000 492,504 (3%) 5,404,360 (37%) 6,082,450 (41%)
Flickr (Total: 36.3m)
14,000,000 12,000,000 10,000,000
2,689,388 (18%)
12,885,979 (36%) 10,082,500 (28%)
8,000,000 6,000,000 4,000,000 4,041,077 (11%) 5,097,200 (14%) 2,838,073 (8%) 1,316,597 (4%)
BY & BY-SA
BY-ND
BY-NC & BYNC-SA
BY-NC-ND
2,000,000 BY BY-SA
BY-ND
BY-NC
BY-NC-SA
BY-NC-ND
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
7
Slide 8: Key observations
• • Greatly varying estimates of size of total CC content pool However, backlink search with both Yahoo and Google yields an almost identical license mix. In this mix:
– 70% of the licenses allow non-commercial use only (NC) – Share-Alike (SA) also a very popular attribute, present in over 50% of CC-licensed items (though SA is anyhow self-propagating) – 25% of the licenses include the ND restriction
•
Generally, two groups of content visible, with one group being licensed under clearly more liberal terms and the other under more restrictive terms BY-ND unpopular in all measurements, although many items licensed under BY-NC-ND; various interpretations possible
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
•
June 14, 2007
8
Slide 9: Reconciling Flickr and search data
Observations • Flickr claims to host 36 million CC-licensed items • According to YBL search results the total CC pool is 37 million items • Flickr appears to host the bulk of CC content • Flickr license distribution is U-shaped vs. bimodal distribution of YBL/GBL/YCC (possibly because photographers license differently) Question • How many more CC-licensed items must there be outside Flickr for the Flickr data to be consistent with the search data? • The solution to a simple linear optimization problem gives that there must be at least 25,500,000 CC-licensed items outside Flickr!
Grand total: 60+ million CC-licensed items online
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
June 14, 2007
9
Slide 10: Volume Generic vs. Jurisdictions
• 80% generic (unported), 20% jurisdiction-specific licenses • Generic historically the only license • Jurisdiction-specific expected to grow at least as fast as generic • “Long tail” is 8 million items, non-negligible
30,000,000
YBL
25,000,000
20,000,000
15,000,000
Generic: 79% (29,195,778)
10,000,000
5,000,000
Jurisdiction-specific: 21% (7,875,039)
Generic Spain France Germany S. Korea Italy Japan Canada UK: England & Wales Poland China (Mainland) Brazil Taiwan Belgium Netherlands Mexico Hungary Croatia Chile Austria Australia Argentina Sweden Israel UK: Scotland Portugal Bulgaria South Africa Colombia Peru Denmark Finland Malaysia Slovenia Malta
10
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
Slide 11: Volume per jurisdiction
2,000,000 1,800,000 1,600,000 1,400,000 1,200,000 1,000,000 800,000 600,000 400,000 200,000 Spain France Germany S. Korea Italy Japan Canada UK Poland China (Mainland) Brazil Taiwan Belgium Netherlands Mexico Hungary Croatia Chile Austria Australia Argentina Sweden Israel Portugal Bulgaria South Africa Colombia Peru Denmark Finland Malaysia Slovenia Malta
YBL
Note: Jurisdiction volume should not be hastily interpreted as country volume since several users may be using the unported licenses or those of another jurisdiction!
60,000 50,000 40,000
Highly correlated
GBL
30,000 20,000 10,000
Spain France Germany Italy UK Canada Japan Brazil Netherlands Belgium Argentina Australia Chile Hungary Mexico S. Korea Austria China (Mainland) Croatia Israel Taiwan Bulgaria Poland Peru Portugal Sweden Finland South Africa Malaysia Colombia Slovenia Denmark Malta
Note: Date of introduction of CC in jurisdiction not taken into account Note: UK jurisdictions grouped together in this chart
-
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
11
Slide 12: Volume per 1000 inhabitants
45.00 40.00 35.00 30.00
YBL
25.00 20.00 15.00 10.00 5.00 -
Note: Jurisdiction volume should not be hastily interpreted as country volume since several users may be using the unported licenses or those of another jurisdiction!
Spain S. Korea Croatia France Belgium Italy Germany Canada Netherlands Hungary Austria Taiwan Poland Sweden Israel UK Chile Japan Slovenia Australia Bulgaria Denmark Finland Portugal Argentina Brazil Mexico Peru Malaysia Malta S. Africa Colombia China
1.40 1.20 1.00
Highly correlated
0.80
GBL
0.60 0.40 0.20
Note: Date of introduction of CC in jurisdiction not taken into account Note: UK jurisdictions grouped together in this chart
Spain France Canada Croatia Belgium Netherlands Hungary Italy Austria Germany UK Israel Bulgaria Chile Australia Finland Slovenia Sweden Portugal Argentina Taiwan S. Korea Malta Japan Denmark Peru Poland Brazil Mexico S. Africa Malaysia Colombia China
-
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
12
Slide 13: License mix per jurisdiction
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Generic Spain France Germany S. Korea Italy Japan Canada UK: England & Wales Poland China (Mainland) Brazil Taiwan Belgium Netherlands Mexico Hungary Croatia Chile Austria Australia Argentina Sweden Israel UK: Scotland Portugal Bulgaria South Africa Colombia Peru Denmark Finland Malaysia Slovenia Malta BY-NC-ND BY-NC-SA BY-NC BY-ND BY-SA BY
• Significant variations, cause unclear • Careful interpretation needed (jurisdictions ≠ countries, also very different “sample size”-volume)
YBL
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
13
Slide 14: Liberal vs. restrictive licensing
• In order to simplify the picture, we can group the 6 licenses into 3 categories: liberal (BY & BY-SA), moderate (BY-ND & BY-NC), and restrictive (BY-NC-SA & BY-NC-ND) Then we can sort all jurisdictions according to their relative use of liberal licenses Yahoo and Google numbers are not so highly correlated for the license mix per jurisdiction as they are for license volume (in other words, they “agree” more on the number of licensed items per jurisdiction than on the license mix per jurisdiction) However, since our analysis suggests that Yahoo data is more complete, we will use YBL here to compare jurisdictions
• •
•
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
14
Slide 15: License mix per jurisdiction (sorted)
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%
Israel Croatia Sweden Bulgaria Colombia South Africa Finland Spain Brazil Generic Japan Portugal Canada UK: England & Wales Netherlands Mexico Denmark Chile Australia Malaysia China (Mainland) Germany Argentina Austria Italy Hungary S. Korea Belgium Poland France Slovenia UK: Scotland Taiwan Malta Peru
100,000,000
58%
10,000,000 1,000,000 100,000 10,000 1,000
30%
100
YBL
% Restrictive % Moderate % Liberal No. of Licenses
• Clear preference for restrictive • Significant variation, but consistent dislike for moderate licenses • Jurisdictions with >100k items use >50% restrictive licensing
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
15
Slide 16: Freedom ratings to capture “mood”
Proposed license ratings
License Creative Freedom Commercial Freedom Total (Mixed) BY 6 6 12 BY-SA 4 5 9 BY-ND 2 4 6 BY-NC 5 3 8 BY-NC-SA 3 2 5 BY-NC-ND 1 1 2
Methodology • Each license is given a freedom rating • Each jurisdiction is given a rating based on the relative popularity of each license in this jurisdiction • Optional adjustment for jurisdiction relative volume, to account for the jurisdiction’s total contribution to the CC content pool
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
16
Slide 17: Uses of ratings
• The willingness of the entire CC author population to license their content under more liberal or more restrictive terms can be summarized in just one number, e.g., according to YBL: 6.21 (out of 12)
Freedom rating Generic - YBL Generic - GBL All - YBL All - GBL Commercial 3.38 3.18 3.27 3.19 Creative 3.06 2.89 2.94 2.89 Mixed 6.44 6.07 6.21 6.08
•
•
Is 6.21 good or bad? Neither, at best what it shows is that the combined effect of the two CC licensing poles (the liberal and the conservative pole) is a rather balanced CC movement, sitting halfway between “all rights reserved” (copyright law) and “no rights reserved” (public domain) Interesting is the fact that the commercial freedom values are higher than the creative values. This is because of the popularity of the SA and ND attributes which have a more negative impact on creative freedom than on commercial freedom (according to our definitions)
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
17
June 14, 2007
Slide 18: Jurisdiction ratings
• • Tables of jurisdiction ratings can be easily constructed for all jurisdictions Jurisdiction ratings should not be hastily interpreted as country ratings!
– after all, 80% of the content is under the generic licenses, and this is not only US-based content – but ratings are useful as the only global indicator we can automatically construct to assess the willingness of authors in a jurisdiction to license their content under more liberal or more restrictive terms
•
Tracking these ratings along with volume data per jurisdiction will allow for some form of measurement of the adoption of the ported licenses in the future
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
18
Slide 19: Creative freedom ratings (max=6)
Position 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Creative Sweden Bulgaria South Africa Finland Spain Israel Generic Brazil Colombia Japan Canada UK: Scotland Croatia Portugal Poland UK: England & Wales Argentina Chile Rating 4.2 4.1 3.8 3.7 3.6 3.6 3.4 3.4 3.4 3.3 3.3 3.3 3.3 3.1 3.1 3.0 3.0 2.9 Position 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Creative Mexico Netherlands Germany Hungary Australia China (Mainland) Austria Malaysia Peru Belgium France Italy Denmark Slovenia S. Korea Taiwan Malta 2.9 2.9 2.9 2.9 2.8 2.8 2.8 2.7 2.6 2.4 2.3 2.2 2.1 2.1 1.9 1.9 1.6
YBL
Rating
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
19
Slide 20: Commercial freedom (max=6)
Position 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Commercial Israel Sweden Croatia Bulgaria Colombia South Africa Finland Spain Brazil Generic Japan Canada Portugal UK: England & Wales Mexico Netherlands Chile Argentina Rating 4.3 4.1 3.9 3.9 3.7 3.4 3.3 3.2 3.1 3.1 3.0 2.9 2.8 2.8 2.5 2.5 2.5 2.4 Position 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Commercial Australia Germany Poland Malaysia China (Mainland) Hungary UK: Scotland Austria Denmark Malta Belgium France Peru Italy S. Korea Slovenia Taiwan Rating 2.4 2.4 2.4 2.3 2.3 2.3 2.3 2.2 2.1 2.0 1.9 1.9 1.9 1.9 1.7 1.7 1.5
YBL
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
20
Slide 21: Mixed index (max=12)
Position 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Mixed Sweden Bulgaria Israel South Africa Croatia Colombia Finland Spain Brazil Generic Japan Canada Portugal UK: England & Wales UK: Scotland Mexico Argentina Netherlands Rating 8.4 8.0 7.9 7.3 7.2 7.1 7.1 6.8 6.5 6.4 6.4 6.2 5.9 5.8 5.6 5.5 5.5 5.4 Position 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Mixed Poland Chile Germany Australia Hungary China (Mainland) Malaysia Austria Peru Belgium France Denmark Italy Slovenia S. Korea Malta Taiwan Rating 5.4 5.4 5.3 5.2 5.2 5.1 5.1 5.0 4.5 4.3 4.2 4.2 4.1 3.8 3.7 3.6 3.4
YBL
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
21
Slide 22: Volume-adjusted mixed index
Position 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Mixed Sweden Spain Bulgaria Israel South Africa Croatia Colombia Finland Japan Generic Brazil Canada UK: England & Wales Portugal Germany UK: Scotland Poland Mexico Rating 8.2 8.2 7.8 7.7 7.1 7.1 6.9 6.9 6.5 6.4 6.4 6.3 5.8 5.8 5.8 5.4 5.4 5.4 Position 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Mixed Netherlands Argentina Chile Australia China (Mainland) Hungary Malaysia Austria France Peru Italy Belgium Denmark S. Korea Slovenia Malta Taiwan Rating 5.4 5.3 5.3 5.1 5.1 5.1 4.9 4.9 4.6 4.4 4.3 4.3 4.1 3.9 3.6 3.5 3.4
YBL
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
22
Slide 23: Looking for relationships…
The differences in the license mix between jurisdictions appear to be unrelated to common economic productivity, political freedom, telecommunications or other national indicators (tested for software piracy level, GDP p.c., unemployment, internet subscribers, broadband penetration, and political, economic and press freedom ratings). Likely the online communities CC users are active in are the most important determinant of the way they license their content. But we do observe that… 1. Google and Yahoo jurisdiction data are positively correlated, with volume data per jurisdiction being more strongly correlated than license mix 2. CC has been propelled forward mostly by developed countries with economic, political and press freedom 3. If we examine the top countries in terms of GDP p.c. then only for those countries CC adoption is positively correlated with piracy rates (further study required)
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
June 14, 2007
23
Slide 24: Conclusions on CC
• License mix • • • Authors prefer the most liberal and most restrictive licenses, moderate licenses neglected Restrictive licenses significantly more popular than liberal licenses (even if CC users presumably choose CC because they find Copyright Law too restrictive) License choice may also depend on the medium type, the community and even the type of content within a medium (ongoing work on these issues) Jurisdiction-specific licenses exhibit significant variation from the usage mix of the Generic license The total CC content pool is at least 40-60 million items An anti-copyright/pro-piracy attitude may be a strong contributing factor for the growth of CC in some developed economies Belonging to a network/community is probably much more important than belonging to a jurisdiction/country
Volume
• •
Overall
•
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
24
Slide 25: Observations on measuring CC
• Even if we could arrive at some conclusions, the data exhibits significant variations depending on the day of measurement and/or the choice of method Search engine results are relatively unreliable for measurement purposes… …however by combining several bad measurements we may get a good result! Better metadata and proper implementation of CC licensing and search capabilities by search engines and key online communities will be essential for tracking the progress and use of CC
•
•
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
25
Slide 26: If you wish to know more about the study: giorgos@smu.edu.sg
Thanks to Ankit Guglani, Giri Kumar Tayi, Warren Chik, Anil Samtani, Mike Linksvayer and Lawrence Lessig who helped with producing and/or disseminating this report Also many thanks to the great folks at the iCommons Summit for their feedback and support
This work is supported by SMU Research Grant 06-C220-SMU-007
June 14, 2007
Data presented herein was collected in early 2007. It is based on (imprecise) search engine estimates and is therefore only indicative of the real quantities whose size we are attempting to assess.
26
