Last Monday, Google deployed a redesign of Reader.
The Internet was not happy. There are a bunch of problems with the new design in my opinion. The design takes up too much space with headers, not enough space for content, and forces anyone who wants to share to use Google+, jumping out of “the flow”. People are rebelling, even going so far as to build a clone of the previous version and offer their services to fix it. Aside from the problems with the new design, though, Google’s Reader has issues, and they were broken far before Monday.

The biggest gripe that I have with Reader is the performance. For what looks like a basic page, it’s remarkably easy to overload your browser with it. I regularly use Reader to browse through thousands of articles at once. Because of the infinite scrolling feature, the reading pane can get very long, and even the best browser can start cranking the CPU cycles. With the old design, I would reflexively hit the “R” button (to reload the page) every hundred items or so. If I didn’t, it would freeze up and sometimes crash the browser.

Not that it’s much better in the new design either - performance is at least a hundred times worse than it was before. In addition to the infinite scrolling, they decided to add a bunch of other new features. The page itself pegs one of my CPU cores with less than 15 stories on the scroll list, and reloading easily can take half a minute if I don’t catch the slowdown fast. The worst offender seems to be related directly to the new Google+ sharing feature. Every story has a “plus one” button attached to it, and takes a round trip from the browser to plusone.google.com to find out how many other surfers pressed the little colorful button. For the company which invented SPDY, there is something wrong with expecting a browser to round-trip hundreds of times just for a glorified counter. I added the domain to the blackhole in my computer’s /etc/hosts file and the new design got faster immediately. I’m still reloading easily every 50 stories though in order to make sure that the browser doesn’t crash.

I’m no javascript application expert, but it seems like the major problem is the length of the reading pane. One solution which I think could be tried is to remove the large selection of ‘read’ stories from the page completely. It is a method that Cullect used when it was in service, and it worked well. It was user-triggered then, but I think it might be possible to make the removal of stories that you are far past automatically in the same way that stories are added automatically just slightly before you get to them on the page, making the infinite length page a little less infinite in length, and making the performance a little more tolerable. For bonus points, make it possible for me to have only one story visible at a time, and load the next or previous story on demand when I use the magic “j” or “k” buttons. After all, the browser is going round-trip to Google’s servers already.