Category Archives: Google

Switching to Google Reader

I held out on using a RSS reader for more than two years. I had a little thingy built so Ping-O-Matic would use its own results and to store the last updated time of every blog it saw, and then I reordered my blogroll based on that. (The original version in WordPress parsed the whole changes.xml from a few times an hour. Remember when that was feasible?) Then I would go from the top of the list to the last one I remember reading, opening each in tabs, and closing as I finished. Incidentally, this kept my blogroll from withering as many do.

At some point I started begrudgingly using Bloglines in 2004, and fell in love shortly thereafter. Where I used to follow a few dozen blogs, now I could consume hundreds and its UI just made sense to me, and just got better over time. I sang its praises often in interviews, mentioning it as one of the handful of websites I used daily. Any outages or performance problems it had seemed minor to me, only reminding me of how much I appreciated the service when it was up. Bloglines search was and continues to be the cleanest in the space.

I don’t recall exactly when many of my friends started switching to Google Reader. I had tried it at its first release and was pretty unimpressed. I saw the hubabub around Gears and how Reader was the first to embrace that. I watched with envy as friends used their trends feature to see nifty data about how and when they read feeds, and use it to cull out the non-essentials, much like I used to with my blogroll.

A week ago, frustrated to no end by a bug in Bloglines telling me to reread things I already had, I decided to make the switch cold turkey. I packed my OPML file and went along to Google Reader full-time. It hasn’t been painless — the keyboard shortcuts are a little funky on dvorak — but it feels better to be on a platform that whether real or perceived feels like it has momentum. That self-fufilling X-factor in apps is one of the magic elements for me. Also as I’m adapting to the UI I feel a lot more efficient than I used to in going through things.

Finally I’ve started reducing my subscriptions, down to 346 now, and I hope to be under 200 within a month or so. I’m thinking of adopted a fixed-number, say 150, and if I want to add one I need to remove another one first so the total is always the same. (I’ve considered this for social networking sites, too.) However it’s probably focusing on the wrong metrics, unread items is more important than total feeds.

MySQL Camp Google Notes

After speaking at Yahoo earlier, I drove a few miles down the street to Google for MySQL Camp. I caught the last session of the day, by Googlers saying how they used MySQL internal to Google. (I assume for the Adwords application.) Here are the stream-of-talking notes I took. The most fascinating bits I took out of it is how they take a partitioning/sharding strategy similar (but notably different in some ways) to and that they use DNS to manage all load balancing, high availability, datacenter failover, etc. DNS is a pretty powerful building block.

Continue reading MySQL Camp Google Notes

The Most Frustrating Thing

It’s Friday, so I’m going to take a few minutes to describe the most annoying concept I see espoused by 95%+ of blogs I read, people I talk to, and friends I have. The problem is until you let go, you’ll never be able to build something truly useful to a non-trivial audience.

Technology doesn’t matter. Design doesn’t matter.

There, I said it.

Continue reading The Most Frustrating Thing

Blogger Rebuilt

“Then, when someone wants to see any of the pages on your blog, those pages are created for them dynamically, on the fly.” Sounds familiar… The new Blogger doesn’t make users wait on rebuilding anymore, nice upgrade! (What happened to all those folks saying static was the only way to scale?) That and their other new features show a real respect and sensitivity to their users, the only thing missing is an exporter. Rebuilding is so 2004.

Banned from Google

Something really weird happened when I had the password problem last week — I completely disappeared from Google. It’s not just the search for Matt, but almost every page on my site has disappeared even for super-obvious searches. This happened within a day of the guy getting into my blog account. I have two theories, one is that when all my links started pointing to blogspot (he changed my siteurl) that triggered some sort of anti-spam flag, and my second theory is that [H]e turned on the new Blog Privacy feature in WP that adds noindex,nofollow to the header of your page, and Google was crawling me that very instant and removed my site. BTW, as an update to the previous entry, I have since found out that I did not have a super-obvious password, but rather he found it embedded in the source code in the SVN repository of a new project I’m working on that hasn’t been released yet. I’ve axed the repo, but at least now I don’t feel so bad about having an awful password on my blog. Regardless, the event was a good excuse to review my password strategy and make sure everything was fairly locked down. Update: I’m certain it was the noindex thing, which looks like it was on for about a week. Let’s see how long it takes to bring everything back and if I rank the same. Update 2: Everything is back to normal. 🙂

Markup Survey

Ian Hixie at Google just published a really awesome web authoring survey of a billion documents. What I found most interesting about reading it was places that things I’ve worked on, notably WordPress and GMPG, popped up.

HTTP Headers — “A pretty significant number of pages include an X-Pingback header (more than the number of pages with the Set-Cookie2 header). In fact, X-Pingback was the 30th most-seen header in our data sample.”

WordPress is one of the few platforms that supports pingback, an alternative to Trackback with a real spec. Apparently there are enough WP pages in the world for this to make a blip on the radar.

Page Headers — “It turns out that a tiny but measurable number of people do use the profileattribute, though. The three most-often used values are,, and This makes XFN the most popular HTML metadata profile!”

Too cool for words. 🙂 Both of these profiles are included by default in some WordPress templates.

rel="pingback" and rel="bookmark" both skirt the charts in the link relationship page. No XFN values made the cut there.

The <a> element — “external seems to be mainly propagated by WordPress, but people have long been asking for a way to label their links as being external vs internal.”

Nice to get a direct mention there, and we’ve supported bookmark and tag from the beginning. All in all the report is a very interesting read, and kudos to Google for doing this type of research and sharing it with the web. I hope to see more of these in the future, it delights my inner markup geek.

GBase and Wikipedia

I’ve seen a lot of people suggest with Google Base they’re competing with Craig’s List and eBay, but I see it much more as a play against Wikipedia. When I chatted with Jimmy Wales he said one of their biggest problems was for things that are relatively structured like the wiki dictionary or quotes having hte data in a totally unstructured wiki format makes it hard to work with. He suggested they would go the direction of mediawiki templates and microformats, which is great to hear. Google has come at it from the other direction, giving you a free-form DB to pour your heart into. I guess the question is, will people become as passionate about distribution (what Google promises) rather than working for the greater good of humanity through the Wikipedia.

Google Analytics

Google Analytics is what I’ve been saying a search engine should have done years ago, provide hosted stats to any website that uses it. People love stats so this will get huge uptake, and probably provide Google really invaluable information. A year or so ago I hoped Technorati would do this for blogs, but they probably saw it as outside their core business. Interesting note: this is the first stat tracking javascript I’ve seen that validates as XHTML Strict. Good job, guys. (Most examples leave the language attribute.)