Netflix Inc. May Have Found the Magic Bullet to Kill Comcast's Connection Fees

Source: The official BitTorrent Facebook page.

Netflix (NFLX +0.01%) may have found a workaround for "the end of network neutrality." For now, the company is paying Comcast (CMCSA 0.01%) special fees to unlock a high-speed Netflix connection to Comcast customers. But in the long run, those massive bandwidth needs could get some help from other Netflix users nearby -- and there's not much Comcast and friends can do about it.

One possible solution involves borrowing technology from movie and music pirates. In a recent blog post, BitTorrent CEO Eric Klinker suggested that Netflix could avoid the drama and costs of striking delivery deals with Internet service providers by mixing in some of his file sharing technology.

Sharing Netflix content the same way a movie pirate might do it could be effective, says Klinker, because "each piece of content ... would have a unique address, and would be streamed peer-to-peer. That means that Netflix traffic would no longer be coming from one or two places that are easy to block. Instead, it would be coming from everywhere, all at once; from addresses that were not easily identified as Netflix addresses -- from addresses all across the Internet."

"To the ISP, they are simply zeroes and ones," Klinker continued. "All equal."

Taking that thought to the next level, tech blog Ars Technica looked through Netflix job postings with an eye toward positions with a peer-to-peer flavor. Sure enough, Netflix is looking for experts in the field.

The company is looking for a senior software engineer with at least five years of peer-to-peer development experience. This person will explore delivery methods for Netflix's Open Connect delivery network, with a special focus on automating and fool-proofing the peer-to-peer component. 

Now, the presence of a single job posting doesn't necessarily mean that Netflix is going all-in on this idea. Like all innovators, Netflix explores new ideas all the time without any guarantees that the latest experiment will ever become a full-on standard. The company might hire an expert, explore the idea for a while, and then dismiss it as unworkable.

Moreover, Netflix is hedging its bets. Ars Technica was told, in no uncertain terms, that Netflix is looking at all kinds of content delivery routes. And the company is hiring engineers and network architects to design "the next generation of the Netflix Content Delivery platform (Netflix Open Connect)."

That said, I think Netflix could be on to something good here.

What are we talking about?
Peer-to-peer file sharing is a proven technology, and it has the power to saturate even a very fat Internet connection under the right circumstances. You don't need a single ultra-fat connection to one well-endowed server. Dozens, hundreds, or even thousands of hookups with lower-speed sources will do just fine.

For example, I'm using a fairly high-end fiber-optic connection capable of sucking down online data at 50 megabits per second. This morning, I used BitTorrent to download the latest version of Ubuntu Linux -- not from a file library on a single server, but via more than 4,600 fellow Linux enthusiasts. The 964 megabytes of operating system goodness arrived in less than four minutes, for an average download speed of 34 megabits per second. There were times when this download stream totally saturated and maxed out my fairly beefy fiber connection. And if I had a faster fiber subscription, I'm sure it would have hit the ceiling of that pipe as well.

There was a time when BitTorrent and other peer-to-peer networks were under the gun, because of their popularity for sharing professionally produced content in illegal channels. Comcast, for one, was known to throttle download speeds of torrent connections.

But that was the start of public outcry to save the open Internet, and the FCC stepped in with new regulations in favor of network neutrality. Under the new framework, it's much harder to justify throttling of any particular network connections, including torrent traffic.

ISPs still try to limit illegal data streams and other inconvenient high-bandwidth connections. But it's much harder now, and only 14% of American torrent connections experience throttled speed limits today.

How could this work for Netflix?
That's why peer-to-peer traffic could sidestep any attempt by service providers or middlemen to limit Netflix streaming speeds. Couple the highly distributed delivery model with strong encryption, which I'm sure that Netflix's content providers will appreciate anyhow, and it becomes almost impossible to target Netflix streams for unusual traffic management practices -- including lower speed limits.

There are many potential ways to take this idea into the living room.

Netflix could set its peer-to-peer component up to only suck down video data from your country (pretty much required), your state (shorter traffic routes, more efficient transfer), your metro area (see above, taken to the next level), or even the neighborhood you call home.

Shortening the path will limit the number of available copies for each movie or TV show episode you're trying to access, but also puts less stress on long-range networks and the Internet backbone. Limiting your sources to just the local neighborhood might only be feasible for extremely popular content (or not at all), but the data would never need to hop from your local connection hub to, say, Comcast's main network at all.

There are also many possible ways to manage the peer-to-peer data library.

All of them require at least some permanent storage on the devices we use for consuming Netflix streams. Whether it's a Blu-ray player, a gaming console, a PC, or a tablet, the device can't pass the buck and start sharing content if there's nowhere to store it.

That rules out plenty of today's most popular streaming devices. Hardware builders who want to boast full-fledged Netflix support in this scenario would have to put some Flash memory or a small hard drive inside connected TV sets or Blu-ray players. Storage-less dongles like the Chromecast can't play this game at all.

But if you have some storage space, your device could download and then share whatever you happen to be watching personally. The downloads could expire when you're done watching, or Netflix could keep a backlog of the last few things you saw. Either way, this would automatically boost overall streaming speeds in lockstep with each show's current popularity.

Or Netflix could manage the source materials, doling out content to user devices in anticipation of what it believes will be popular enough to deserve a boost.

Or maybe the company would just spread its entire library across user devices, as evenly as possible. That would take out all the guesswork and automation, and simply increase access speeds for everything.

Has Netflix found a way to reduce or eliminate these annoying rebuffering screens?

Combine these strategies any way you like, and add in other possibilities that didn't occur to me, and you'll get a near-infinite number of final designs. In every single case, Netflix would then bake in the traditional method of sucking down video streams from centralized servers and regular content delivery networks. I would argue that a hybrid system like that would see very few "Buffering" or "Loading, please wait" pauses in the final user experience.

If Netflix can build all of this without incurring legal headaches and the wrath of nervous content providers, peer-to-peer networking may very well be the magic bullet that saves network neutrality. And that would be a huge win for Netflix.