/ 9 December 2009

Streaming will never stop downloading

Someone convinced the record and movie and TV industries that there is way of letting someone listen to audio or watch video over the internet without making a copy. They call this “streaming” audio, and compare it to radio, and contrast it with “downloading”, which they compare to buying a CD.

The idea that you can show someone a movie over the internet without making a copy has got lots of people in policy circles excited, since it seems to “solve the copyright problem”. If services such as Hulu, Last.fm and YouTube can “play you a file” instead of “sending you a file”, then we’re safely back in the pre-Napster era. You can sell subscriptions to on-demand streaming, and be sure that your subscribers will never stop paying, since they don’t own their favourite entertainment and will have to stump up in order to play it again.

There’s only one problem: Streaming doesn’t exist.

Oh, OK. Streaming exists. It is a subset of downloading, which comes in many flavours. Downloading is what happens when one computer (a server, say) sends another computer (your PC, say) a file. Some downloads happen over http, the protocol on which the web is based. Some happen over BitTorrent, which pulls the file from many different locations, in no particular order, and reassembles it on your side. Some downloads take place over secure protocols like SSH and SSL, and some are part of intelligent systems that, for example, keep your computer in sync with an encrypted remote backup.

Streaming describes a collection of downloading techniques in which the file is generally sent sequentially, so that it can be displayed before it is fully downloaded. Some streams are open-ended (like the video stream coming off your CCTV camera, which isn’t a finite file, but rather continues to transmit for as long as the CCTV is up and running).

Some travel over UDP, a cousin of the more familiar TCP, in which reliability can be traded off for speed. Some streaming servers can communicate with the downloading software and dynamically adjust the stream to compensate for poor network conditions.

And of course, some streaming software throws away the bits after it finishes downloading them, rather than storing them on the hard-drive.

It’s this last part that has the technologically naive excited. They assume that because a downloading client can be designed in such a way that it doesn’t save the file, no “copy” is being made. They assume that this is the technical equivalent of “showing” someone a movie instead of “giving them a copy” of it.

But the reason some download clients discards the bits is because the programmer chose not to save them. Designing a competing client that doesn’t throw away the bits — one that “makes a copy” — is trivial.

All streaming involves making a copy, and saving the copy just isn’t hard.

Does this matter? After all, if the entertainment industry can be bought off with some pretty stories about a magical kind of download that doesn’t make a copy, shouldn’t we just leave them to their illusions?

What harm could come from that?

Plenty, I fear. First of all, while streaming music from Last.fm is a great way to listen to music you haven’t discovered yet, there’s no reason to believe that people will lose the urge to collect music.

Indeed, the record industry seems to have forgotten the lesson of 70 years’ worth of radio: people who hear songs they like often go on to acquire those songs for their personal collections. It’s amazing to hear record industry executives deny that this will be the case, especially given that this was the dominant sales strategy for their industry for most of a century. Collecting is easier than it has ever been: you can store more music in less space and organise it more readily than ever before.

People will go on using streaming services, of course. They may even pay for them. But people will also go on downloading. Streaming won’t decrease downloading. If streaming is successful — that is, if it succeeds in making music more important to more people — then downloading will increase too. With that increase will come a concomitant increase in Big Content’s attacks on the privacy and due process rights of internet users, which, these days, is pretty much everyone.

If you want to solve the “downloading problem” you can’t do it by waving your hands and declaring that a totally speculative, historically unprecedented shift in user behaviour — less downloading — will spontaneously arise through the good offices of Last.fm.

There are more problems, of course. Streaming is an implausible and inefficient use of wireless bandwidth. Our phones and personal devices can be equipped with all the storage necessary to carry around tens of thousands of songs for just a few pounds, incurring a single cost. By contrast, listening to music as you move around (another factor that has been key to the music industry’s strategy, starting with the in-car eight-track player and continuing through the Walkman and iPod) via streams requires that you use the scarce electromagnetic spectrum that competing users are trying to get their email or web pages over. Count the number of earbuds on the next tube-carriage, airplane or bus you ride, multiply it by 128kbps (for a poor quality audio stream) and imagine that you had to find enough wireless bandwidth to serve them all, without slowing down anyone’s competing net applications. Someday, every 777 might come with a satellite link, but will it provide all 479 passengers with enough bandwidth to play music all the way from London to Sydney?

What’s more, streaming requires that wireless companies be at the centre of our daily cultural lives. These are the same wireless companies that presently screw us in every conceivable way: charging a premium for dialling an 0870 number; having limits on “unlimited” data plans; charging extra for “long distance” text messages. They’re the same wireless companies whose hold-queues, deceptive multi-year contracts, surprise bills, and flaky network coverage have caused more bad days than any other modern industry.

Why would we voluntarily increase our reliance on expensive, scarce wireless bandwidth delivered by abusive thugs when we are awash in cheap, commodity storage that grows cheaper every day and which we can buy from hundreds of manufacturers and thousands of retailers?

Especially when every streaming song creates a raft of privacy disclosures — your location, your taste, even the people who may be near you and when you’re near them — that are far more controllable when you listen to your own music collection.

Finally, there’s the cost of going along with the gag. The more we pretend that there is a technical possibility of designing a downloader that can’t save its files, the more incentive we create for legal and technological systems that attempt to make this come true. The way you hinder a downloader from saving files is by obfuscating its design and by creating legal penalties for users who open up the programs they use and try to improve them. You can’t ever have a free/open source downloader that satisfies the desire to enforce deletion of the file on receipt, because all it would take to remove this stricture is to modify the code.

An incentive to obfuscate code, to prohibit third-party modifications and improvements, and to weld the bonnet shut on all the world’s computers won’t actually stop downloading. But it will have anti-competitive effects, it will reduce privacy, and it will slow down innovation, by giving incumbents the right to control new entrants into the market.

Hard problems can’t be solved with technical denialism. The market has spoken: people want to download their music (and sometimes they want to stream it, too). The supposedly for-profit record labels could offer all-you-can-download packages that captured the law-abiding downloader, and then they could retain those customers by continuing to make new, great music available. It’s been 10 years since Napster, and the record industry’s hypothesis that an all-you-can-download regime can’t work because users will download every song and then unsubscribe from the service is not borne out by evidence. The fact is that most downloaders find cheap, low-risk music discovery to be a tremendous incentive to more consumption, as they discover new music, new artists, new songs and new genres that tickle their fancies.

Selling customers what they desire is fundamental to any successful business. If Big Content can’t figure out how to do that, then we can only pray for their hasty demise, before they can do too much more damage to humanity’s most amazing and wonderful invention: the internet. – guardian.co.uk