Computers and Internet

The End of the Beginning


The Netflix and Comcast deal marks the birth of a brand new unregulated futures trading market trading in an emergent and still undefined currency called “network efficiency”. The impact of this on the web will be profound and more complex than what we may realize if we simply argue over the end of network neutrality. Good will come from this (companies will be better able to provide the best experiences for their users) while there will also be harm (eg pure market approaches can reinforce monopolies). 

Full Article

Comcast and Netflix have come to a landmark agreement whereby Netflix gives Comcast money and Comcast ensures that Netflix sucks less for Comcast subscribers. For a great description of this deal check out Timothy Lee’s article on Wa Post here. Tim does a very good job of describing the immediate and short-term issues this agreement introduces.

I could decry this simply as the 3rd and final major milestone in the inexorable death of net neutrality – and actually I think that it is. But that doesn’t say much about what the future will really hold, which is what I’m interested in.

How things kinda worked

The internet still works, right? OMFG IS TEH INTWERWEBZ BORKED? If you can read this then the interwebz are not totally borked.

Yesterday, bits on the internet were all created kinda sorta equal. (Not entirely – copyrighted bits are subject to DMCA take-downs, some countries block some bits in the name of federalism, child pornography bits live entirely on Darknets, etc…) But basically TCP/IP – the protocol that forms the spine for the internet today – is designed to treat all packets the same. It may deliver them out of order or unpredictably but it does that to all of them and generally speaking it delivers them all, albeit sometimes via some remarkably confusing routes.

Yesterday there wasn’t a lot I could do to make my bits jump ahead of your bits on the internet. I could use an independent CDN (Content Distribution Network) as a way to push copies of my bits out to optimized caches on the internet but in the end those networks still are part of the “all packets are created equal” internet dating back to DARPA. Yesterday CDNs represented a relatively small specialized market where I could spent money to give my packets a specialized advantage. There really was no effective way for me to spend money to guarantee that all of my future internet packets will perform substantially better all the time for a large number of my users.

Funnily enough, the inverse of this wasn’t true – while I couldn’t spend money for special treatment, other people could spend money (or refuse to spend money) so that my packets would be treated badly. Large network companies do this by throttling certain packets and this was ruled as entirely okay by the FCC in the US. Comcast has the technical ability to make packets go to the back of the line every time but until today I couldn’t pay them to make my packets go to the head of the line no matter how much money I wanted to spend (short of buying Comcast outright).

What changed?

Over time network efficiency has been relatively fixed while capacity is subject to significant growth. Capacity grows as investments are made in things which provide more of it (eg Google’s fiber network, 4G build-outs, etc.). Over time efficiency changes slowly because the network has specs and needs to be compatible. The specs are basically as efficient as they are. (Network smarty pants will no doubt pick at this but I think that we can probably agree on the high-level view that it’s a lot harder to change raw network efficiency than it is to change raw network capacity.)

If you are a network provider (like Comcast) and you have growth in demand for more packets you have to grow your capacity because you have no way of radically changing efficiency. In fact from a network operator perspective if may seem like your network becomes less efficient over time because demand for content is outstripping your capacity. In what might seem like a terribly ironic twist of fate the packets you deliver to your paying subscribers might be undercutting your own biz model because those subscribers have decided to pay Netflix instead of you for access to content that you used to sell them access to.

Meanwhile if you are a subscriber-funded services company on the network but you have no network of your own (like Netflix) your entire value proposition is predicated upon your paying users getting a great experience from your service on other people’s networks. Perhaps unsurprisingly you also have a highly vested interest in both network efficiency and capacity. You can improve efficiency on your side of the problem space (this is unarguably something that Netflix spends a huge % of their intellectual horsepower on) but you don’t own the network so you can’t do much to affect efficiency or capacity once your packets leave AWS and enter the network.

If you are a Comcast subscriber and you are reading this post and at the exact same time someone else in your house is watching a Netflix movie, and they are both routed into your house via Comcast, then this blog post might load slower so that the packets which are part of the Netflix movie arrive faster. Netflix pays Comcast to get their streaming packets ahead of my WordPress blog packets when we are both serving Comcast subscribers.

Okay, so what does that really mean?

Someone can now trade something of value to someone else for value in return that they couldn’t get before. Netflix (someone) is trading money (something of value) to Comcast (someone else) in exchange for network efficiency (value in return).

That’s it? That’s all that happened? That doesn’t sound very revolutionary, does it? It doesn’t sound it but I think that it’s actually vastly more disruptive when you think about it that way, not less. Why? Because until today network efficiency wasn’t really trade-able. Things that are tradeable form markets. They behave like commodities or currencies.

The implications of this are HUGE. New markets grow unpredictably and emergent behaviors define the future as they go. Weird shit happens in new markets. I think we are now looking at a future network efficiency market where companies spend money to ensure specific behaviors for the packets they care about.

A hypothetical example

Someone can now trade something of value to someone else for value in return that they couldn’t get before.

I’ll pick an obvious hypothetical example for how this might work in the future by looking at a someone called Apple.

Something of value? Apple has money – lots of it. As of today they have more than $150B in pure, spendable cash. They have more money than anyone else. More money than god. Network operators want money and they can essentially manufacture efficiency from thin air to make up for the competitive pressures they are feeling elsewhere. It’s a perfect deal!

Someone else? Comcast, Telcos, any/every one providing internet connectivity to a substantial number of users.

Value in return? Imagine a future where Apple sees value in efficiency guarantees for packets reaching an Apple device. Apple could be quite specific and use this as a way to ensure the best experiences for their users and 3rd-parties who fully embrace the iOS ecosystem by only providing efficiency guarantees to iOS apps and native Apple SW & services or they could generalize it to all packets going in and out of their devices. They may not be able to pay to have Google packets slowed down but if they put enough money into a futures trading market they can ensure that their packets are treated the best way possible, and in a futures market there are always winners.

Result? Apple buys network efficiency for their devices. Apple users get demonstrably better experiences on Apple devices. Apple enhances their extremely valuable brand. Apple users are happy. The people negatively impacted aren’t Apple customers and won’t really know to blame Apple when something works less efficiently on a competitive device. Network ops make money. The network, as a whole, no longer behaves efficiently but instead is pseudo-managed as some packets are sometimes treated as more equal than others on some parts of the network. Some packets will get screwed and will always wind up at the back of the line.

Network effects 

I’m not calling out Apple because I have any inkling of them planning on doing this and they aren’t bad guys in this narrative. I’m using them as an example of one possible future that is clearly plausible but perhaps not immediately obvious in a world where network efficiency is something anyone with enough “something of value” can acquire. Money isn’t the only tradeable currency here – imagine a crossover between the Personal Data Economy and the Network Efficiency Economy where packets that support big data analytics are more efficient than those that are encrypted and/or behaviorally masked. You can see an easy Google play here – they may not have as much money as Apple but they can counter Apple cash investments with what they do have in ads, analytics and networks.

I gotta point out that from my perspective neither Comcast nor Netflix are bad guys here either. It’s more complicated than that. What’s becoming clear to me is that we are at the dawning of a brand new unregulated futures trading market trading in an emergent and still undefined currency. The impact of this on the web will be profound and in many cases more complex than what we get if we simply say this marks the end of network neutrality. Good will likely come from this (companies can secure the best experiences for their users, we could have guaranteed QOS for first responders in a disaster, …) while there will also probably be bad (bullying, perpetuation of network effects by dominant players, …).

Prospective build outs and new technology will rush in to capitalize on new value. SW defined networks must become the norm as network HW providers must respond dynamically to participate in the new market themselves. Networks could choose to reserve QOS super-futures for use in an emergency by first responders to ensure connectivity…

As we established earlier efficiency and capacity are basically fixed at any given point in time so if someone can outbid everyone else for Network Efficiency Futures (NEF) then by necessity when that future becomes the present their packets will be ahead and the rest of the packets will be in line behind them. Because new markets have a tendency to make secondary and tertiary instruments tradeable you can imagine hedging and pure futures investments by non-service-providing 3rd-parties as well. These could be used to effectively block the best efficiency for others. Counter allocation of network efficiency – spending money to make some packets perform poorly – can become a real thing and that probably means it will.

I don’t think this all adds up to the end of the web, but I do think it is the end to its beginning. Yesterday network efficiency wasn’t a currency and tomorrow it will be.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s