Commentary: It's Not AV1 vs. HEVC
With Apple joining the Alliance for Open Media (AOM) earlier this year, there has been some interesting discussion and some fairly robust blogging over the past couple weeks about what this will mean for the industry.
And I am glad about that. It is certainly an interesting development and one worthy of news, debate and discussion.
But one thing concerns me. From the great article by my fellow Streaming Media writer Jan Ozer to a blog from Beamr and another from Bitmovin, there is still a tendency to write about codec advances in terms of "alternatives" or "A vs. B."
Any of us who have been in the industry for more than a couple of decades will be accustomed to this line of thinking. I have penned numerous articles covering technology advances framed in the context of "Windows Media vs. Flash" or "MP3 vs. Real Audio" or "MSS vs HLS" and so on. As commentators and journalists we do love a good polarization of opinion.
Often we have industry ties and subconsciously want to enhance our own commercial positioning in one side of that argument or another (even if ethics decree we declare that interest). So we make strong arguments, and the inevitable outcome is that we talk in terms of extrapolated futures that show why one side will outlast the other, or why we think potential consumers of one technology or another will succeed and others will fail.
This is a perfectly natural legacy from an era where technology changes were costly, complex, and involved a lot of risk analysis and change management. Every investment in a technical strategy traditionally required a large CAPEX investment, and mistakes could make dramatic differences to OPEX margins. But gone are the days when each additional format means a new hardware stack and physical infrastructure capable of doing that and nothing else.
When a strategy relied on Windows Media, we would invest in many Windows licenses. We knew that the reach would be limited to the reach of Windows-licensed technology. We knew that if the competition invested in Adobe Flash Media Server, they may have a different cost dynamic—and if they hit the demographic for the content better than Windows Media, then they would succeed where we would fail, and we would additionally have to bear the costs of adding the Adobe platform to our own technology mix just to stay in the game.
And in the past few days reading the blogs about Apple joining AOM you can see this type of reasoning throughout all the commentary. There is rarely a black-and-white "one will win and the other will fail" posturing, but there is a lot of "one is safe and the other is risky" talk at the moment.
But, just as in 2007 when I started to evangelize about moving video encoding workflows into the virtual hosting environment, and in 2010 as my company started to educate the sector on moving CAPEX to OPEX in the cloud, I feel it is time to shake the bag a little.
It's like this: We are no longer in a market which has to make "either/or" decisions. We are now in a market that is agile and adaptive by nature, and no longer has the same CAPEX risks and change management involved in its decision-making. Competition has completely changed. Risk has completely changed.
In tomorrow's (and maybe even today's) increasingly adaptive market we are no longer producer-led, but rather consumer-led. Any company that is building "product" and hoping the market will come can start very small, with almost no risk. This is now absolutely true of all aspects of software. Indeed, if you run a company making "monolith" software products with slow periodic release cycles, then your products' days ARE numbered—and I will debate this as long as it takes for you to accept this! I have a decade of experience to draw on in examples to support my fanaticism.
Obviously, at the moment the hardware market itself is still catching up. But essentially GPU hardware speeds and storage access speeds (which underpin concerns such as performance and power consumption and so on) are the one remaining factor in this space that is just now starting to change. With Intel (and others) introducing GPU into the common CPU, and with FPGA following close behind, if you make a codec choice today you should be able to readily change or add a new codec tomorrow, without having to start again from scratch.
I emphasize the word "add" for a reason. It is incredibly important to understand that the entire ecosystem is becoming more and more flexible. With microservice architecture sweeping the distributed compute design philosophy end-to-end, changes to every atomic stage in every workflow can be adapted on the fly and on demand in response to each and every individual user's specific requirements. There is no longer one-size fits all, or one product for the market.
In fact it grates with me that I still walk the halls of so many exhibitions and conferences were companies are selling their customers "products" that encapsulate a specific set of features, and who only offer variation on a platform-wide, periodic release cycle basis. This is out of date and encumbers the market in a sort of "commercial treacle," meaning that customers are not getting what they want, they are only getting what vendors offer. I have written about my favourite term 'Service Velocity' here.
So in response to those who have framed AV1 as "more exciting than HEVC" or who have conversely framed HEVC as "much more important today than AV1 may ever be" or who are focused on the hardware limitations of one over the other, or who think that Apple's announcement is a confusing endorsement coming so hot on the heels of its endorsement of HEVC, and to those who think that these are "wars" that will have just one winner, or those who think that VP9 is a better investment than H.264 or vice-versa (etc.), you are living in a world that will soon no longer exist.
The reality is that "versus" is becoming a legacy concept. Today you increasingly no longer make "either/or" decisions. Those are decisions made by the market. The real impact of Apple's decision to support HEVC was that overnight there were hundreds of millions of HEVC-capable devices capable of supporting what had, until then, been a niche technology.
Any technology will be successful just so long as the customers can use it when they want to use it. The only things that can fail is you not making it available when a customer wants it, or the licensing of underlying technology not being set at a price point that your operating margins can bear.
So let's make 2018 a year of not talking in terms of one thing versus another and choosing sides, and instead start to focus on how all options can co-exist and be made available. We can let the consumers decide.
The old realities that used to dictate codec adoption no longer apply. Opening up new markets now matters more than reducing operating expenses. How are HEVC, AV1, and VVC positioned for the future?
18 Nov 2019