Compelling Compression or Revolutionary Hype?

O'Reilly editor Nat Torkington pointed out this news story about an allegedly revolutionary compression technique from ZeoSync. Here are my thoughts on the subject:



Maybe I'm jaded. My first impression is that this story is bigger news for mathemeticians and theoreticians (and patent lawyers) than your average geek. Among other unanswered questions is the time (i.e. computational power) required to compress and decompress images and other data. Most compression algorithms are assymetric, meaning that it takes longer to compress data than to decompress it. Fast decompression is essential for real-time streaming. (For example, a Zip file isn't an appropriate format for real-time streaming.)



Within reason, the time required to compress something hasn't been a factor to date. Traditionally, the rendering time for CGI has been substantially greater than compression time.



But any technology that, say, makes it possible to stream large, high-quality videos over a 56K modem, will result in huge server loads as millions of people access it. There might also be a substantial computational cost to compress the data in, say, realtime, apart from transmitting it. It won't be as popular as ripping MP3s if it takes too long on a typical desktop PC. Also music includes a lot less data than typical video. So if it is an order of magnitude better compression, you might see a lot more video swapping, but I wouldn't call it revolutionary.



Given the legal opposition from various copyright holders (RIAA) to digital transmission, I'd be surprised if we see commercial products from any new algorithm in less than a year. Maybe they mean the end of 2003.



I suspect that ZeoSync is less of a crock than Ginger/IT/Segway, which I said from day one was either some kind of fuel cell or ridiculous hype. If anything, it seems ZeoSync has exaggerated how revolutionary any compression improvement will be, but that gives me no reason to suspect that they can't improve on current compression ratios. Lossless compression of video is nice but not mandatory. The kind of data sets for which lossless compression is mandatory (such as numbers and text) tend not to be very large. So unless you run out to buy widescreen HDTVs, I'm not sure you'd care about lossless video.



I don't think your average Joe is ready to pay for compression, per se. It would have to be licensed and built into a service, kind of like VCR+ programming, which adds a few dollars in licensing fees to any VCR you buy. Now, if it incents studios to release movies over the internet before releasing them to theatres, that would be revolutionary.



### Get your Daily Bruce! ###

Will ZeoSync's compression technology revolutionize computing in our lifetimes?