I have a question that is eating me up. I have been reading and learning all sorts of stuff that is waaay above my abilities, and all the while trying to learn the academic language that any answer I might be given will likely be provided in.
I have worked with Content Delivery Networks all my life. I have been deeply interested in IP Multicast. IP Multicast has never realised its potential to be the most efficient way to scalably deliver live video to global audiences. This was largely not for technical reasons, but for 'human' reasons and as the Internet grows, the problem of 'deployment' becomes increasingly more intractable.
This is a loss of opportunity for the streaming video industry, and I have spoken at many conferences about this. The elegant IP Multicast protocols have other applications and where they are deployed I believe they offer a very interesting range of efficiencies for computing novel problems. I say 'novel' because while IP Multicast is little used and little known in the developer community there is little innovation around the technology.
On occasion I have alluded publicly to some 'thought experiments' - wild ideas that arise from being very familiar with a technology in one specific application, but that appear to address a different problem space altogether.
In particular:
Multicast Pre-Compute
I propose that IP Multicast can provide an approach that can enable individual classical computers in a large distributed computing infrastructure to 'pre-compute' answers to problems before they are given any data at all! In effect the overall system learns from itself, at accelerating speed as time progresses.
Quantum Multicast
Further, while practical technologies are many years away, I propose that the combination of (IP)* Multicast and Quantum Computing's emerging capability to address specific problems are well suited 'bedfellows' since if my 'pre-compute' model above is applied in a Quantum world then it will lead to a significant further acceleration in an environment that, today, already promises almost incomprehensible efficiency gains, even without the Multicast 'Pre-Compute' model. It is this convergence of ideas that this blog is named after: Quantum Multicast.
This blog aims to capture these wild, developing ideas, my musings, notes, and attempts to articulate the idea better and more rigourously.
And it needs to be taken apart by you - it must survive scientific rigour. While I may delete pointless abusing profanity on the comments sections, I will try to follow links to recommended research, and to understand comments that are made in response to my blog posts.
I am a layman - not an academic (today, as i start to write the blog!) Perhaps my language will change and my ability to describe the detail will refine into the amazing (yes i said amazing) maths that i have been barely grasping over the past months, in particular as i bring myself up to speed with Quantum Physics, Mechanics and Computing. I have been reaching out to academics I know already, and to others i have found through deeper and deeper reading. I am lucky: My local university (Sussex in the UK) has a leading Quantum Mechanics team. I hope to develop deeper contact with them.
Many years ago I also met Prof. Jon Crowcroft, then at UCL but now at Cambridge. He was one of the guys who led the very early thinking about IP Multicast. I wrote to him about my idea and his hearty response was (in short) "wow. wacky, weird, interesting. Do it". And so here we are. Asking questions...
Comentarios