Friday, July 16, 2004

Daiwa EuroTelcoblog No. 57: Friday, 16th July, 2004 - Some very different angles on on-demand TV
A couple of weeks back I wrote about some takeaways from the Profiting from On Demand TV event sponsored by Informa Media. Basically, my distillation of what I saw and heard led me to a pretty pessimistic conclusion over the likely success of the incumbent carriers in cracking this market. Since then we've had some new data released about shifts in file sharing patterns, and a kind reader has also alerted me to some new developments which potentially make the whole issue of content distribution somewhat more exciting and problematic for everyone. The following tries to weave these two developments into some sort of alternative view of the on-demand TV phenomenon.

Recent data

Firstly, over the past week, two reports on file sharing have been published (by the Motion Picture Association of America and the OECD), and two producers of P2P traffic policy solutions (CacheLogic and BayTSP) also produced some interesting stats.
  • CacheLogic found that as recently as June, global P2P platforms averaged 8m concurrent users at any given time (actually the press release says 10m, but this is a peak usage figure observed over six months). These 8 - 10m concurrent users typically shared a staggering 10 petabytes (that's 10m gigabytes) of data.
  • The MPAA estimated that 2.6bn files are copied each month on P2P platforms globally.
  • The OECD, basing its analysis on data from BigChampagne, estimated nearly 10m concurrent users globally in April 2004, up from just over 7m in April 2003. It also provided a breakdown of estimated users by OECD member state. Europe claimed seven of the top ten slots, with Germany at number two after the US, accounting for 10.2% of all global P2P users.
  • The OECD notes that the fastest growth in share of P2P usage has come from France, Germany, Italy and Japan.
  • In terms of penetration of total population, in France and Germany (joint third place after Canada and the US) the OECD estimates that 0.6% of the total population engages in file sharing. Canada is double that level, but adjusted for broadband penetration levels, some of the European penetration figures are actually higher than that in Canada.
  • Interestingly, while the OECD has found a strong negative correlation between income/education levels and downloading behavior in the US, in France this relationship does not appear to exist. The OECD produced data from ART which appears to show a roughly equal tendency to download irrespective of income, and a greater tendency based on education level. The reading is that, at least in France, file sharing is on the way to becoming a more mainstream and widespread phenomenon over time.

Running beneath all this data is an apparent shift in both P2P platform preference and the content being shared. CacheLogic pointed out that reports of declining file sharing activity are based on faulty measurement, as the industry groups such as the RIAA tend to track KaZaA and other Fast Track-based networks, which have gone ex-growth. If the net is widened to include other platforms, such as eDonkey and Bit Torrent, usage is growing and becoming more bandwidth intensive. The OECD data backs up the claim that there has been a platform shift, as it shows users of Fast Track platforms remaining fairly stable at around 4m since late 2002, while the overall number of P2P users nearly doubled.

What is promoting the shift? Video, apparently, particularly in the case of Europe. The OECD found that in Germany, Italy, Belgium and France, video accounted for c.25 - 40% of all files shared on KaZaA, though KaZaA is generally acknowledged as being inferior for swapping large files like video in comparison to Bit Torrent and eDonkey, the two platforms which have seen the largest growth. Bit Torrent in particular has been huge according to CacheLogic, which estimates that it now accounts for 53% of all P2P traffic globally, up from 26% in January 2004. The MPAA study asserts that globally, 24% of internet users have downloaded a movie at least once, with Korea leading the pack at 58%. However, France (27%), the UK (20%), Italy (20%), and Germany (19%) are all in line with the global average and the proportion in the US. The survey also found that 20% of those with no file sharing experience intended to start at some point in the future.

Some of this data is bound to be a bit skewed, because Bit Torrent is often used also for distribution of large software files, which may suggest that some items are being misclassified as movies. However, there are still some fairly serious implications for both the carriers and anyone involved in content ownership/distribution:
  • For the carriers, the shift from music files (average file size 4MB) to video files (ranging from 100MB to upwards of 600MB in the case of a full-length film) implies a lot of network capacity is being soaked up in the process. CacheLogic's presentation has some interesting graphics, and makes the point that in one extreme case, it determined that 30% of the total P2P traffic on the network of one large ISP was generated by the availability of a single 600MB file. With P2P traffic already estimated by some to account for >80% of ISP traffic, this dramatic uplift in bandwidth intensity is a pretty alarming prospect. CacheLogic's own estimate of transit costs to ISPs in Western Europe associated with P2P for this year is EUR100m.
  • For the content owners and distributors, the piracy issue is again intensified, but perhaps with more impact in the film industry as opposed to music, due to its more capital-intensive nature. This, in turn, may bring in more litigants via suits such as the Belgian case against Tiscali as an "enabler."
  • For carriers looking to build a business case for TV/VOD as an application on their DSL platforms, we have to question what sort of dilution impact they may see from an embedded and expanding user base more intensively sharing video content.


Against the background of all these data points, a kind reader this week alerted me to the existence of something very interesting called Torrentocracy. I spoke briefly with the developer of this application, which was just released last month. He was very quick to point out that he is keen to avoid any identification of his innovation with illegal file sharing, and there is a very clear policy statement on the website:

"Torrentocracy is not about railing against or stealing from big media. Instead it is about creating this all access network from the bottom up in the best interest of consumers. Just imagine how dull the internet would be if all the content was controlled and doled out by just a few large corporations. Now realize that this is how TV has always been. If the revolution will not be televised, it will instead be brought to you via torrentocracy."

My inclusion of it in this piece is purely my own decision as an illustration of some of the innovations coming out of the open source developer community which challenge old assumptions about content creation and distribution. Depending on the uses to which these innovations are put by the individual end user, the picture painted in the first part of this piece could grow more complicated.

Torrentocracy ( is a fusion of Bit Torrent and RSS, and there are others such as Buttress and Azureas which are perhaps better known (Azureas claims 10m downloads to date), but where Torrentocracy has moved forward is in the integration of this with the television. This was achieved via integration with the MythTV ( open source PVR system, a phenomenon we have been looking at from time to time over the past six months. Torrentocracy, however, seems to revolutionize the promise of such a system, by allowing syndication of content between large numbers of individuals using RSS feeds with torrent files embedded. Theoretically, we are looking at a new concept distribution platform for a variety of audio and video content, including entirely non-corporate, independently produced material. Obviously, being open source, such a system could be put to other uses not envisaged or sanctioned by the developer. My main interest is in the fact that Torrentocracy and any subsequent similar developments are clearly an interesting step in decentralizing control of content creation and flow from the global media conglomerates, and may prove an additional source of pressure on commercial on-demand offerings, whether from the telcos or the established players in cable/satellite.

Is the industry watching the right channels?

One potential inhibitor of such a platform may be the diffusion of Linux operating systems. Statistics based on system registrations at Linux Counter ( rather unhelpfully estimate somewhere between 2.9m and 72m users of Linux worldwide, so there is a wide margin of error here. It is interesting to note, in passing, that the country breakdowns of Linux registrations in the site are heavily weighted towards Europe. However, at this point, the main takeaway in my mind is that the current speculation about winners and losers (ADSL, cable, satellite) in the on-demand video content world may get considerably more complicated as a result of this sort of innovation being driven by independent developers working in relative obscurity.

No comments: