Do it with modern encoders and modern settings, and back it up with proof from actual ABX testing and not unsubstantiated boasting. Then we'll have something to talk about. Until then it's dreams and excuses.

I don't mind people doing irrational things. I use lossless encoding as much as I can too. It makes me feel good to know that I can have bit-for-bit accuracy of reproduction. But if you think your ears are actually good enough to sense bit-for-bit accuracy, you're delusional, and reckless crowing about your magical ears does nothing but perpetuate the mythology used by predatory "hi-fi" businesses to fleece neophytes and people who should know better. If you want to believe in that stuff, fine, but unscientific bullshit is exactly what enables bad vendors to sell gold-plated HDMI cables at a huge markup and bass-increasing green permanent markers to suckers who think they can hear quality differences that simply do not exist.

I haven't tried that. Is it really much better with other codecs or uncompressed music though? Seems like if you don't have the vocals in a separate track, you're going to have to try to remove them algorithmically, which is going to be tricky but shouldn't be worse with Mp3 than any other format. If all the tracks are flattened into one, they're flattened into one regardless of the encoding.

Speakers have different efficiency levels. Better speakers are more efficient and need less amperage to create the same volume of sound.

Nice. I have the RF-62 II's. I love Klipsch too.

Nice jab! Would be better if it were relevant though: we're discussing 40KHz dynamic range and MP3 sounding shittier than uncompressed, which simply isn't true (unless you're using a very old encoder or very low quality settings).

To each his own. I fatigue more rapidly with the headphones than with the stereo.

I mostly agree, but I think two good speakers make music sound better than 5 shitty ones.

Yeah, fuck science and facts of human biology!

Buy 2 servers, preferably used storage array servers. Start a raid 5 or 6 array on one. (This server is the storage server.) This is your main storage drive. Store ALL data on it. It's helpful to have it support multiple access methods (SMB, NFS, iSCSI, etc). You could go full OS like Debian or something like OpenNas or OpenFiler (BSD based).

On the second box, add as much storage as is accesable on the first. This is the backup server. Run a cron job to regularly r-sync the data off the Storage server over to the backup server.

In this configuration, you have some redundancy in the RAID and a true backup in the second server. You also have the ability (hopefully) to drop in drives as you need so you can expand as you go. And if the hardware it's self breaks, you can simply replace it and keep going.

That makes the technical argument even weaker.

The counter-techincal argument is that those users already don't get h.264 on XP. So what's the difference between not having it because the browser doesn't let you use the system libraries and not having it because there are no system libraries? As presented, the difference appears to be that you aren't really getting the same browser on different OSes if there are dependencies on your OS and OS version.

I think the technical retort there doesn't hold a lot of water. After all, your OS probably came with a browser that isn't Mozilla-based which will gladly use native libraries for this kind of thing. Moreover, whether or not they decide to do this, the amount of work developers have to do to support the ridiculous WebM format alongside H.264 isn't going to change: you'll still have to have your content encoded twice and you'll still have to sniff out which version to show.

I think if you frame the argument as "why aren't we doing this?" instead of "why should we do this?" it becomes a lot more clear which course of action is the right one: the one that means a better experience for your users, which means better OS and hardware integration and better battery life when using your browser. Users plural may care about consistency, but a single user is much more interested in features and performance.

promising that you will never have to worry about losing or moving them across devices ever again.

You know, until they shut down the service in 18 months because it isn't popular enough...

One thing I remember very clearly from my college physics course was an explanation of how going faster than the speed of light leads to causality violations. I suppose it's possible that causality isn't something we "need" but it's going to be very hard to have a sensible concept of physics if "A causes B" has to be thrown out the window. I find the probability that ONE instrument doing ONE experiment that shows that maybe something went faster than light a lot less likely to be true than "A causes B" being true and that ONE experiment simply being bad.

I quite agree with you about strings, but FTL is completely insane.