>It appears you didn't really read the article. There's no need to pull 100gb of data. I read the article, paying special attention to section 4, aptly titled

>It appears you didn't really read the article. There's no need to pull 100gb of data. I read the article, paying special attention to section 4, aptly titled "Smoke and Mirrors". There's always a need to pull in as much data as the user wants to actually interact with. No matter what tricks you pull, you'd still need about 1s to pull in a 10MB photo over a 100mbps connection at full speed, and most people don't have that. Reading data off a disk is an order (or two, with an SSD) of magnitude faster. And sometimes, I just want to flip through the library and look at a lot of photos without having to wait seconds for each one to load. Sure, the layout algorithms are still important, and that's what a lot of the article is about. But all king's horses and all king's men can't make the latency of a javascript UI with a network backend comparable to a compiled program interacting with a disk. Thank you, but smoke and mirrors aren't for everyone. Google Photos fulfills an obvious need, but it's not a Picasa replacement.