The hype about Joost and the reality

22. Mai 2007

Joost Logo

There haven been a lot of Bablefish and Google Translate requests so I translated my German Post „Der Joost-Hype und die Realität.“ There might be some quircks but I hope the message is clear.

More than half a year ago I received my first invitation for Joost, which at that time still operated under the project title The Venice Project. Since that time Joost achieved an impressive attention for this application both in blogs as well as in traditional media.

Joost Posts in Blogs

The extreme buzz even established Joost-invitations as a currency in blogs. While complex experiments by readers were required during the good old days of the old Venice Project, the price has now gone down to a simple comment. Despite this fact invitations can still generate significant transactions (+6000 comments).

There is one common agreement between (most) bloggers and media: Joost revolutionizes the TV and the founders of Skype and Kazaa are doing their next big thing. Indeed Janus Friis and Niklas Zennström could turn out to be the main factors of success for Joost and its investors, because that’s the main advantage compared to the competition, which does more or less about the same.

Joost also benefits from the fact that approach is easy to communicate. Media does understand what Joost does: It is simply “broadcast” via the Internet. Joost offers DRM, Geo-Targeting, Anti-Piracy-Measures, targeted advertising and the complete environment can be controlled. Media houses do not have to compete “against” User-Generated-Content, but can simply store their complete archive and wait for advertising money to be generated. But even this advantage and the two founders, who are still working mainly for Ebay and therefore have to leave the implementation to Dirk-Willem van Gulik and Fredrik de Wahl, cannot conceal some serious problems.

The Annoying Thing

First there are a few non-substantial things at Joost (which I personally don’t like) which probably do not pose bigger problems in extending the offering.

As Spiegel Online nicely puts it: Doubleclick, Username, Password and the show begins. My experience was mainly: Doubleclick, Crash, Download of a new version, setup Screen Name, enter even more data … wait … show. Even after half a year of beta testing crashes are the rule rather than the exception. The update policy with the immediate expiration of old versions and the growing appetite for more personal data remind me a bit of old Spy-/Adware stories.

The Joost-(View-)impression is additionally hindered by two main circumstances. At first, with each user action (change channel, jump episode …) advertisement is popping up. Generally nothing wrong with that, but the advertisement happens even I had only watched the channel for a second and it generally takes three seconds for the ads to load, followed by five seconds of advertisement itself and again another three to five seconds for the next channel to start. Imagine a splash screen every time you change channels while zapping on traditional TV or changing an URL on the web – unthinkable.

The second annoying thing while watching anything are the extreme fluctuations in image quality and – in my opinion even worse – the stop and go if the stream cannot be received at the same speed.

Even more annoying for the future buyer might be something totally different. Considering the recent financing round of $45 million there are only a handful of usual suspects who come to mind for the sale of Joost (Amazon, Ebay, Yahoo!, Microsoft, Viacom, Disney, News.Corp …). One of these giants will have to generously relieve the current investors of their problems, because Joost does not have any substantial assets to justify the current valuation. The core technology, the Global Index is not part of Joost and was not part of the Skype- or Kazaa-deals. This technology is still owned by Joltid from Janus Friis and Niklas Zennström and just waits to be re-used in the next big project (P2P based MPOG??) after the Joost-exit. The other technologies used for Joost are mainly based on Open-Source-components. And as the founders are not part of the management of Joost, the current investors‘ only the chance for a successful exit is an acquisition.

The problems

These obstacles are overall only larger or smaller annoyances, which can either be turned off or which won’t have implications on the future development of Joost. However, Joost also has to fight with more serious problems, which are inherent in the given architecture and which cannot be overcome easily.

Desktop-application

The first of such is he fact that Joost is a desktop application. Therefore the hurdle for installation is given. This will be followed by the fact, that Joost cannot participate on any well known network effect of the browser based Internet. Links to selected shows are only possible via detours, direct links, widgets or embeds are not possible. In an era where more and more applications are available as internet based versions (Google Office …) and the lightnets contribute a substantial amount to the revenues and successes of the last years, Joost wants to move TV back into the darknets.

This won’t work for the masses. Joost at this time largely profits from „Filesharing-Switchers“, who give up P2P applications in favor of Joost. This will only work until the geographical limitations are fully operational. Then they will return to P2P applications in order not to miss the US-start of any popular show.

Beside the problems of Joost being a desktop application, the hardware requirements are another challenge. First, a typical simple DSL (1 Mbit/s) connection is not sufficient and second, the minimum hardware requirements are really exactly the minimum and more likely to make the application crash than work. On many PCs and Macs Joost simply cannot be installed and many firewalls in companies will block Joosts traffic.

P2P-Streaming is not efficient

Joost is based on a Peer-to-Peer-technology. It has been widely communicated that P2P-TV is more cost efficient and more effective in delivery, because no centralized servers are required and the users are covering the bandwidth costs. This statement is wrong. Peer-to-Peer-Streaming in contrast to e.g. Bittorrent does not add different data slices into one single file, but instead Streaming always needs a specific part of the data in order to continue the stream. This reduces the amount of available peers drastically and leads to dropped frames and a poor Quality of Service if one peer quits Joost or gets interrupted. In addition the available upload bandwidth is much lower than the download bandwidth.

Joost tries to solve these problems in three different ways (this pdf source was pulled after Joost realiced it contained internals) (1) there are centralized servers which are seeding all contents, (2) these centralized servers are also responsible for delivering the long tail contents, which are only watched from a few peers and therefore cannot be delivered efficiently over a P2P-network, and (3) the Joost servers are trying to close the DSL-Gap by substituting the missing bandwidths. Add to these assumptions the fact that P2P-Streaming generates 40% overhead, resulting in additional CPU power for the peers (because of needed recalculation and reconfiguration of the source files) and additional bandwidth for double sent packages, and you will reach the conclusion that P2P-Streaming is badly suited for the delivery of video contents (Joost employees aren’t allowed to comment on the P2P-Overhead, therefore hard facts are missing).

Therefore it is no surprise that the server went down with each new opening of another Beta-Test, as they must handle the main load of the traffic. (Since my first writing I was informed that not the Long Tail Servers went down but the Channel Metadata Servers.) Investigating such setup more in-depth you come to the conclusion that Joost would be better served with a centralized or de-centralized (CDN) server architecture than P2P-Streaming:

  • A large amount of the traffic is generated in the long tail, which is handled by Joost-Servers.
  • 66% of the other traffic the Joost-Server must carry because of the DSL-Gap. Per peer and hour: 320 MB down 105 MB up = 215 MB Gap, this assumes that Users close Joost after they are done watching, why should they keep Joost open?
  • Add seeding and overhead and we come to the conclusion that the servers in one way or the other need to handle and carry the main load of traffic anyway – with or without P2P.

This problem could lead to massive quality reductions once Joost is available for everybody, an analogy to the worse quality at Skype after public launch.

P2P-Streaming is expensive

Due to the fact that every peer is forwarding contents it uses more bandwidth than it should. That’s the idea behind Joost: the traffic-bill for the service provider (Joost) shall be as low as possible. Apart from my opinion that this equation will not work out (see above), Joost needs tremendous bandwidth on the consumer side. Assuming a daily Joost-consumption equalling traditional TV consumption, the end-user needs 1 GB of traffic per day just for Joost (320 MB down + 105 up * 2,x hours). These are tremendous amounts, which could disturb the mixed calculation of the ISPs dramatically. This could lead into either the fact, that the bandwidths of poweruser get limited during peak hours, at any time the ports get blocked, usage of Joost gets excluded in T&C or the pricing for all users needs to go up. At the end of the day the end users pay in one way or the other to help saving money for the corporates.

Pretty interesting to see at this point, that the big media houses honor the bandwidths provided by the endusers in their P2P offerings and even pay for it in various cases. Joost in contrary offers nothing comparable, but earns money with advertisement distributed with the end users‘ bandwidth. In the end Joost heats up the net-neutrality debate once more.

All contents must be shipped to Leiden

In order to integrate videos or shows at Joost, they must first be shipped to Leiden in the Netherlands. There they will get transcoded and distributed to the Joost-servers. For this process there might be a better approach in the future, but for the legal negotiations upfront there is none. Copyright holders cannot just sign up at Joost and if they could, they would usually not have worldwide rights for such contents. Joost would almost need unlimited financial and human resources to negotiate all licenses fast to establish a compelling program. So far I experience humble results. Now we even see, that based on demand from copyright holders artificial boundaries had to be established to ensure that contents licensed for the US cannot be watched in Germany and the other way around as well. Until Joost has overcome such hurdles and can offer more than some rare popular contents, a few years will pass by.

Joost is neither Fish nor Fowl

Joost wants to combine the Best in TV and from the Internet. The result can be compared by watching their application and a TV optimized interface:

Joost Interface EyeTV Interface

Joost obviously couldn’t decide what is the Best from TV and what is the Best from the Internet. Therefore they choose picture quality and full screen size (TV) as well as On Demand and Chat (Internet). The result is a hybrid, which neither suits the demands from Internet nor from TV. For a Lean-Back-Usage (TV) the interface is not suited, as it cannot be read nor navigated from a one meter distance, unless you own a 42 inch Plasma-Screen ;-) The on-demand features are easily available and in higher quality with any DVR or available cable services.

A Lean-Forward-Usage (Internet) of Joost is hindered due to the obtrusive full screen mode and the desktop application. Joost does not integrate well at all in the known web environment. The user shall not use Joost in context of websites and his normal surfing behaviours on the web.

Conclusion

The main advantage of Joost are the founders, who guaranteed with their former successes a lot of publicity, trust and connections. But the affinity of the founders to P2P-technology could turn out to be a critical disadvantage, as the latter is not necessarily suited to meet the goal.

Therefore the founders at the end of the day are no insurance that the Joost model will work – perhaps only for the investors, because an exit is pretty likely. One of the big ones will buy Joost sooner or later. Most likely Yahoo! – they’ve already bought Broadcast.com and Joost would fit into the picture quite nicely ;-)

Verwandte Beiträge

Weitere Fragen?

Haben Sie Fragen zu meiner Tätigkeit oder benötigen meine Unterstützung?

info@gugelproductions.de