|
Tech Law Journal |
 |
| News, records, and analysis of legislation, litigation, and regulation affecting the computer, internet, communications and information technology sectors | |
|
|
|
| |||||
|
Editor's Notes: BRIAN L. BUCKLEY, SBN 116705 UNITED STATES DISTRICT COURT
1. My name is Richard L. Stout. In my capacity as an author of computer and Internet-related books and articles, and as an Internet web traffic consultant, I am known professionally as "Rick Stout." I am over 18 years of age and am not a party to this action. 2. Unless the context indicates otherwise, I have personal and first hand knowledge of the facts set forth herein, and, if called as a witness, I could and would competently testify thereto. 3. I am the author of the book Web Site Stats: Tracking Hits and Analyzing Traffic (published by Osborne McGraw-Hill in 1997). For this book, I exhaustively researched web server software, traffic tracking techniques and tools, and analysis methodologies and software. In this book, I provide detailed analyses of more than a dozen of the most popular free and commercial server log analysis software packages and services as well as ad campaign management software such as RealMedia’s Open AdStream product. 4. I have also authored articles on the subject of Internet web site "traffic data" for popular print and online publications. My article, "Tracking Down Your Web Site Traffic" appeared in the October 1996 issue of NetGuide magazine. Another of my articles related to Internet web site traffic, titled "Web Server Log Files" was published in the online magazine, Mainspring.com, in November of 1997. Professional Background 5. I graduated from National University in San Diego in 1980 with a Bachelor’s Degree in Business Administration. In 1985, I accepted an accounting position in the Accounting department of Western Health Plans, Inc., a major Southern California health maintenance organization. Simultaneously, I enrolled in an Accounting curriculum, continuing my education and completing a second Bachelor's degree, in Accounting, in 1986. I was then manager of the Accounting Department of Western Health Plans and Controller for Western’s largest subsidiary, the Greater San Diego Health Plan. 6. During my employment with Western Health Plans and the Greater San Diego Health Plan, I began my study of computers and computer software and developed a system within the Finance department to facilitate general operational systems and basic general ledger accounting. I also studied for the Certified Public Accountant’s examination. I passed the CPA exam in 1988 and accepted a position with the accounting firm of Ernst & Whinney (the predecessor to Ernst & Young). 7. Because of my knowledge of computers and software, and my experience as Controller for the Greater San Diego Health Plan, Ernst & Whinney assigned me directly into the firm’s Consulting Services Group rather than the more traditional Audit or Tax departments. In the Consulting Service Group, I concentrated mostly on computer and network installations and software implementations for some of Ernst & Whinney's clients. Ernst & Whinney subsequently became Ernst & Young, where I remained employed. 8. While with Ernst & Young, I began working with Internet-related software applications. I implemented a Unix server in the downtown San Diego office of Ernst & Young, and a world-wide Internet-based email system for the San Diego office of Ernst & Young. 9. In 1990, I left Ernst & Young to pursue a software-development opportunity with a San Diego-based firm called Home Healthcare Systems, Inc.. Home Healthcare Systems developed software for the home healthcare industry. 10. In October of 1991, I left Home Healthcare Systems to pursue freelance consulting and writing opportunities which had arisen. During 1991 and 1992, I provided consulting services to numerous clients and set up and implemented Unix servers, networking, and associated software. 11. In 1992, I authored my first book, Power Shortcuts for Quattro Pro for Windows, which was published by MIS Press. 12. In 1993, I co-authored another book, The Internet Complete Reference, with a well-known Internet author, Harley Hahn. This book was published by Osborne McGraw-Hill in 1994. The Internet Complete Reference was an easy-to-read guide to using the Internet. At the time, the Internet was more difficult to use than it is today. In 1994, the Internet was largely an esoteric command-line-driven Unix-heavy network most heavily trafficked by "scientific" and "academic" types of individuals. 13. Also in 1993, Mr. Hahn and I began work on The Internet Yellow Pages which was subsequently published by Osborne McGraw-Hill in 1994. Both The Internet Complete Reference and The Internet Yellow Pages reached best-seller status on several well-known national lists including the New York Times' Bestseller List. The Internet Complete Reference was reviewed in the New York Times on June 5, 1994. The Internet Yellow Pages was reviewed by The Wall Street Journal on July 18, 1994. 14. Also during 1993, I began work on a book for Glencoe, the textbook division of Macmillan/McGraw-Hill. The book, The Peter Norton Introduction to Computers, was published by Macmillan/McGraw-Hill in 1995. 15. In 1996, I authored a spin-off to The Internet Yellow Pages titled The Internet Science, Research & Technology Yellow Pages. In addition, I authored The World Wide Web Complete Reference. a technical book on the World Wide Web. While writing this book, I researched the most popular "web server" software, the methodologies web servers employ to record web site traffic data and statistics, and the software available for analyzing web server "log files." 16. In late 1996, I completed the book Web Site Stats: Tracking Hits and Analyzing Traffic (published by Osborne McGraw-Hill in 1997). For this book, I exhaustively researched web server software, traffic tracking techniques and tools, and analysis methodologies and software. I provided detailed analyses of more than a dozen of the most popular free and commercial server log analysis software packages and services as well as ad campaign management software such as RealMedia’s Open AdStream product. 17. Although my book, Web Site Stats: Tracking Hits and Analyzing Traffic, has been in print for several years, a current search of the amazon.com and barnesandnoble.com web sites publications related to "web traffic" reveals that Web Site Stats: Tracking Hits and Analyzing Traffic is still the dominant technical guide to collecting and analyzing web site traffic data. My Work on the Within Lawsuit 18. On or about July 6, 1999, I was formally retained by the Law Offices of Brian L. Buckley, counsel for defendants in the within action, in the capacity of consultant and possible expert witness in the area of Internet web site traffic data and statistics. Defendants' counsel informed me generally of the issues in this case and of the of the need to bring evidence to bear on the "effect on the market" element of the "fair use" defense. Defendants' counsel stated to me that they had obtained my book, Web Site Stats: Tracking Hits and Analyzing Traffic, which Defendants' counsel said had been helpful in formulating Defendants' discovery requests propounded to Plaintiffs. 19. When I was retained by Defendants, Defendants' counsel informed me that my role, initially, would be to advise Defendants concerning what computer data would typically be collected on Internet web sites similar to Plaintiffs' Internet web sites, washingtonpost.com and latimes.com, which could show the effect, if any, of the operation of Defendants' web site, freerepublic.com, upon web site traffic to Plaintiffs' web sites. 20. In addition, Defendants' counsel informed me that I was to assist Defendants with the technical aspects of resolving discovery disputes with Plaintiffs over discovery requests propounded by Defendants for information and documents related to web site traffic data and statistics. Defendants' counsel informed me that, after such data was produced by Plaintiffs, I would then analyze the data to determine what conclusions can be drawn from the data, if any, and I would prepare the data in a form acceptable to the Court. 21. I initially informed Defendants' counsel that it was indeed true that, depending upon the exact configuration of the "web server" software installed on Plaintiffs' web sites, data may be collected by Plaintiffs which could show the number of web site "hits" which were "referred" to Plaintiffs' web sites from Defendants' web site within a given time period. I further informed Defendants' counsel that -- depending upon what software was installed on Plaintiffs' sites -- it may be possible to analyze the activity of those "referred" users during their visits to Plaintiffs' site. I informed Defendants' counsel that, for the most complete, accurate and thorough analysis of the web site traffic data, the best and most complete evidence would be to obtain all of the web server "raw" logs from Plaintiffs' web sites for the relevant period of this lawsuit. I informed Defendants' counsel that, using these "raw" logs, I could analyze the raw data and prepare summary reports and statistics from same. I further informed Defendants' counsel that in absence of the ability to obtain the raw logs -- depending upon what software is implemented on Plaintiffs' web sites to do their own analysis and reporting on the raw logs -- it may still be possible to use Plaintiffs' summary reports to estimate the effect of the operation of Defendants' web site upon the web site traffic on Plaintiffs' web sites. 22. As Defendants' counsel "met and conferred" with Plaintiffs' counsel over discovery disputes, Defendants' counsel provided to me some of the "meet-and-confer" correspondence between the parties. Defendants' counsel expressed considerable frustration to me over what Defendants' counsel characterized as Plaintiffs' resistance in producing the data necessary for my analysis. 23. I was ultimately informed by Defendants' counsel that Plaintiffs contend that it is too burdensome in terms of manpower and hardware capabilities to produce the raw logs for the entire relevant period (from which I could have derived the most accurate and comprehensive data). I was somewhat skeptical of Plaintiffs' position because, in my opinion, based upon the deposition testimony of Washington Post employee Eric Schvimmer, and Los Angeles Times' employee Mark Richards, related to how Plaintiffs' data is stored, Plaintiffs could produce the requested data either by simply copying the tapes on which the data is stored, or by dedicating one computer with a sufficiently large hard disk, and one person to monitor the process, to uncompress the stored data and produce the raw logs for any period requested (for which there is data). [Redacted] Alternatively, it is my professional opinion, based upon Plaintiffs' deposition testimony, that Plaintiffs could use the [Redacted] reporting software installed on their computers to produce new reports from the these logs -- even for the time periods for which Plaintiffs now claim to have produced no reports. 24. Notwithstanding my opinion in this regard, Defendants' counsel informed me that, in a concession to Plaintiffs, they had agreed to try to "make do" with only two weeks of sample raw log data for each of Plaintiffs' web sites, latimes.com and washingtonpost.com, and that I would have to rely, to the extent possible, on the derived summary reports produced by Plaintiffs' for only a subset of the relevant months of this litigation. Such summary reports show only the total figures per day, or per month, for only certain limited variables, and do not show the precise web site activity occurring "dynamically" during the day -- as the raw web server logs would. 25. After being informed by Defendants' counsel of the foregoing concessions to Plaintiffs, I was then informed by Defendants' counsel that Plaintiffs would not be producing the two weeks of data for the washingtonpost.com site after all. Defendants' counsel informed me that Plaintiffs would only be producing the two weeks of "raw logs" for the latimes.com site only. Defendants' counsel informed me that Plaintiffs were either unwilling or unable to produce the two weeks of "raw logs" for the requested time periods for the washingtonpost.com site. Thus, for the washingtonpost.com site, I was informed that I would have to rely for my analysis on Plaintiffs' summary reports only -- summary reports which contained much less information than even the summary reports produced for the latimes.com site. 26. I have proceeded to analyze, to the extent possible, all web site traffic data so far produced by Plaintiffs. I have not yet received, and I am informed and believe that Plaintiffs have not yet produced, the two weeks worth of sample "raw logs" for the latimes.com site already agreed to by Plaintiffs. 27. There are two primary factors to take into account in evaluating the effect of the Free Republic web site on the L.A. Times and Washington Post web sites. These are:
28. With respect to quantifying item 1 (the beneficial effect to the plaintiffs web sites of increases in traffic as a result of referring links on the defendant’s web site), the data provided by plaintiffs is incomplete and sketchy, however it is sufficient to draw some conclusions. 29. With respect to quantifying item 2 (the detrimental effect resulting from loss of traffic to FreeRepublic.com), the defendants have failed to provide the information necessary for a detailed analysis. Nevertheless, I can draw some general conclusions based on the information that is available. Beneficial Effect to Plaintiffs 30. Note 1: This analysis relates to the L.A. Times web site only. The Washington Post provided insufficient data for a similar analysis of that site’s traffic. 31. Note 2: Because most of the data made available for this analysis relates to January 1998 through August of 1999, this analysis is limited to that time period. 32. In order to quantify the effect on the plaintiff’s web sites of referrals from the Free Republic web site, it was necessary to estimate the number of advertisement impressions (or views) on the L.A. Times web site that were the direct result of readers being referred to the L.A. Times web site from the FreeRepublic.com web site. 33. If the plaintiffs had used the full functionality of most traffic analysis products and services for the period in question, estimating the ad views for sessions referred from FreeRepublic.com would be a simple matter. The [Redacted] software that the L.A. Times now uses to track traffic and ad banner impressions summarizes and reports total impressions and total unique viewer sessions by month. 34. Unfortunately, the L.A. Times only began using the [Redacted] software in September of 1998 with the first whole month reported being October of 1998. Therefore I had usable [Redacted] reports only for the period October 1, 1998 through August 31, 1999 (eleven of the twenty months of the analysis period). This data from the [Redacted] reports (Total Impressions and Total Unique Viewer Sessions) is presented on page 4 of the spreadsheet titled "Analysis of LA Times Web Site Traffic"). In the column labeled "D" in that worksheet, the average number of impressions per unique viewer is calculated on a monthly basis. In April of 1999, the number of unique viewer sessions (labeled "(C) Total UVS" on the worksheet) took an unexplained drop from a number consistently above 3 million to a number consistently below 2 million. 35. Although total impressions reported by the [Redacted] software peaked in April 1999 and declined somewhat thereafter, the drop in total impressions was not a significant factor in the drop in unique viewer sessions. The result of the unexplained drop in unique viewer sessions is a significant increase in the critical statistic – the average number of impressions per viewer. In the interest of conservatism, I chose to use only the consistently lower numbers for the period October 1998 through February 1999. The average number of impressions per unique viewer session for this period is 6.5 (see exhibit 1008, page 2, top worksheet). 36. Because the L.A. Times did not have usable reports from the [Redacted] software prior to October 1998, it was necessary to establish a relationship between the total impressions reported by the RealMedia software during the period those numbers were available and the only other statistic that was available for the entire analysis period. This statistic is the total requests serviced by the L.A. Times web server. Total requests differs from total impressions in that total requests includes the raw web server requests for every embedded element of a web page (text and graphics) while total impressions relates to views of an advertisement banner—generally a one to one correlation with page views. 37. In addition to the [Redacted] reporting software, the L.A. Times has also used a popular [Redacted] software package (called [Redacted]) to report statistical information it summarizes from raw web server log files. Analog reports include daily, weekly, and monthly summaries in addition to referral and request reports. The L.A. Times provided monthly reports for the months of January 1998, and April 1998 through December 1998. They provided request reports for the period March 1998 through August 1999. No reports were provided for the month of February 1998. 38. Because neither the monthly reports nor the request reports spanned the entire analysis period, some reconciliation and extrapolation was required.. Refer to Exhibit 1008: For those months for which monthly reports were available, the number of requests reported by the monthly reports was used. For those months for which both monthly reports and request reports were available (May 1998 through December 1998), I performed reconciliation to prove that although the request reports do not report the total number of requests for a period, extrapolating the total number of requests is very accurate. Most of the request reports report the top 5,000 files requested sorted in descending order of the total number of requests for each. At the bottom of the reports are a total for the total requests for the reported files and a percentage indicating the percent of total views that total represents. 39. For example, the July 1998 Request Report indicates that the most requested file for that month was [Redacted]. The next most requested page was [Redacted]. he report continues to the last detail line of the report which reports that the 5,000th most requested page was the page at [Redacted] with 174 requests. The summary line at the end of the report indicates that the pages reported in the report account for a total of [Redacted] requests and it also indicates that this number of requests is 89.46% of the total requests for the period. 40. For each month for which request reports were available (May 1998 through August 1999), I divided the total number of requests reported by the percentage of the totals those numbers of requests represented to extrapolate or "gross up" the number of requests reported to the estimated total requests for that month. 41. For each month for which both request reports and monthly reports were available, I compared the calculated (extrapolated) total number of requests with the actual total number of requests reported by the monthly report. This reconciliation (Exhibit 1008, page 3) shows that for each the months examined, this extrapolation resulted in an error rate of less than .01%. Stated inversely, the accuracy was consistently greater than 99.99%. Based on this reconciliation, I am comfortable relying on this methodology to estimate the total requests for the remaining months of the analysis period accurately (January 1999 through August 1999). Because no reports were made available for February 1998, that month is estimated by averaging the preceding and succeeding months (January 1998 and March 1998). 42. Using the above methods, I estimated the total requests for the months of the analysis period for which reports were not available and derived a table of total requests for the analysis period. The following table sets forth the derived total number of requests for each month of the analysis period: [Redacted] 43. The following table compares the derived total requests and unique viewer sessions for the months for which the unique viewer session quantity is valid (before the unexplained drop in the [Redacted] reports). I used this table to derive a factor representing the average number of requests per unique viewer session. The resulting factor is 30.4.[Redacted] Deriving this factor allowed me to derive total unique viewers for the entire analysis period (See Exhibit 1008, page 1, columns B, D & E). 44. The only referral data for the analysis period made available by the L.A. Times was generated by the [Redacted] software for the months of March 1998 through December 1998 (See Exhibit 1008, page 4, column E). Based on this data, I derived a factor to apply to the remaining months of the analysis period to estimate total referrals per 100,000 requests (See Exhibit 1008, page 2). The factor derived is 3.75 referrals per 100,000 requests. 45. Using actual referral numbers for the months for which referral reports were supplied, and the factor I derived above for the remaining months, I estimated total referrals (Exhibit 1008, page 1, columns G and H). With an estimated total number of referrals for each month of the analysis period, I estimated total referred impressions by multiplying the derived total referrals (column H) by the average number of impressions per unique viewer previously estimated at 6.5 (Exhibit 1008, page 2). The result of this computation is set forth in Exhibit 1008, page 1, column J. 46. Over this analysis period, I estimate that referrals from FreeRepublic.com result in a benefit to the L.A. Times site of over 20,000 ad impressions per month. Based on the L.A. Times published ad rates (using the $ [Redacted] impression rate for campaigns of less than 300,000 impressions), I can quantify this benefit at just under $ [Redacted] per month Detrimental Effect to Plaintiffs 47. In order to perform a detailed analytical analysis of the detrimental effect to the plaintiffs of the FreeRepublic web site, the detail raw web server log files are required. These log files have not been made available. However several conclusions can be drawn nevertheless. 48. First, relative to the L.A. Times web site, the Free Republic web site is trivial in size and traffic. The entire web site is served by a [Redacted] operating system. The entire traffic volume of this site is trivial with respect to the plaintiffs web sites. 49. Second, only a small percentage of the articles on the Free Republic web site are from the plaintiff’s web sites. The detrimental effect to the plaintiff’s web site could not possibly be more than this small percentage of a total traffic, which itself, is trivial with respect to the L.A. Times site. 50. Finally, the general public seeking to read the news has no reason to believe that the Free Republic web site would be a better place to read the news than at the L.A. Times web site. Indeed, most members of the general public seeking to read the news have never even heard of the Free Republic web site Summary 51. In my professional opinion, any detrimental effect to the plaintiff’s web sites resulting from the existence and operations of the Free Republic web site are trivial and probably are more than offset by the beneficial effect of the referrals the Free Republic site feeds to plaintiffs web sites. I declare under penalty of perjury under the laws of the State of California that the foregoing is true and correct. DATED: October 18, 1999 Richard L. Stout
|
|
