Hacker News new | ask | show | jobs
by dodders 4453 days ago
The idea of lossless compression in a smaller file size than anything else on the market spoiled the entire show for me.

Call me picky, but...

9 comments

On the contrary, I appreciated this detail. It's difficult to come up with a plausible-sounding start up idea that doesn't sound like a joke. The lossless compression idea sounds plausible to the mainstream audience. For the hacker audience, it plays on a different level - it's a cue for us to get.
To be fair, perhaps it is lossless, but only for a subset of all possible data files. Like compressive sensing, which beats the Nyquist-Shannon limit, but only if conditions about sparsity are met.

In fact, the Nyquist-Shannon theorem itself is a kind of lossless compression. It says you can potentially throw away data without losing information.

I thought that at first too, but it would actually be a funny plot development if someone later points out that information theory dictates that the claims are impossible. Then the main character would receive more VC funding to pivot into a completely different but even more ridiculous idea.
It was deliberately ambiguous what they were building since it's more or less a MacGuffin but there was some talk of it allowing faster search over compressed content which, to me, sounds like plausibly valuable technology.
I haven't watched the show, but wasn't BWT a lossless compression with a smaller file size than anything else on the market? Or are they talking about comparing lossless to lossy?
That bothered me at first, but remember that it's not the tech that's the point of the show, it's the culture. Who's on First isn't about baseball.
I gotta admit, as soon as they said "and it's lossless!" I was a little disappointed
Lossless compression of non-random data is very possible, though. Video and audio files are far from random, and I don't believe that lossless compression is "a solved problem" (even by the business standard) for those. (However, it is a bit far-fetched that someone would invent a groundbreaking new algorithm in the space for a silly website.) General lossless compression may not exist, but we're talking about highly patterned (low entropy) subspace of all possible files/bit-strings.

Lossless compression requires assumptions about the data, of course, because making some patterns shorter requires making others come out longer. Random data will, in general, get larger under a lossless compression algorithm.

Don't compare it to reality, compare it to The Big Bang Theory.