The gargantuan file was in C:\Documents and Settings\All Users\Application Data\Microsoft\Search\Data\Applications\Windows\Projects\SystemIndex and was similarly named to other files that were 4KB, 1, KB, 171 KB... nothing at all like the 35GB monster.

I deleted it and rebooted. Search still seems to be working fine and now I've got 1/3rd of my hard drive back. However, there's a new file in there. It's now at 6.8 megs and seems to grow by about 1K every 5 seconds.

Very weird. Guess I'll have to keep an eye on it from now on. I'm not too happy to have another thing on my list of stuff I have to remember to do.

Death Rail 2000! :Thmbsup:

So the benefit is only on my own network? It doesn't use anyone else's bandwidth?

I can't see how this scheme would confer any real benefit. The bottleneck in most home networks is the pipe to the ISP. Max out your cable or DSL bandwidth, and it doesn't matter how many machines you've got trying to pull down a stream - only so many bits are going to get through at a time.

No "algorithm" behind the bottleneck is going to be able to alter this equation. Now, if software outside your network were re-compressing and re-streaming, that might help. The biggest benefit I could see would be that multiple peers serving up a stream from different locations would effectively avoid any upstream bottlenecks between you and the video site - swamped servers, a downed router, etc.

I'd be very suspicious about the effectiveness of recompression, though. Video streams are already highly compressed, using asymetrical compression technology that applies the best compression possible for the given content. Any re-compression would have to be done on-the-fly, using a more generic, real-time algorithm. If you saw any benefit at all, it might be 10%-15% - not 300% - 500%.

Compression and streaming are also CPU intensive. I doubt you could run this in the background without seeing a substantial performance hit, provided it's doing what it claims.

This strikes me as very similar to those "RAM doubler" programs that used to be around. Using compression on data in RAM sounded plausible enough to people who didn't really understand the technical issues, but it proved to be nothing but snake oil. I wouldn't be surprised if this was of the same ilk. Improving video downloads is a problem lots of people would like to see solved, but that doesn't mean that a cheap, easy solution to the problem really exists.