As part of my ongoing love affair with Google App Engine, I put together a tool to make an RSS feed out of some of the commenters on Hacker News. It’s at http://ultrasaur1.appspot.com (App Engine limits you to 10 applications, so I’ve had to recycle some).

It’s just a weekend project, but it’s already using 4% of my free quota for only 2 users. Most of that’s parsing the feeds (it’ll end up ignoring 99% of the comments) — but it doesn’t bode well for scalability — and I think I’m ok with that. I did implement one optimization, if two people ask for a feed consisting of exactly the same commenters, it’ll be cached.

What are the odds that many people will want to follow exactly the same commenters? Pretty high if we assume most users will never actually customize their list (maybe higher if the default list is any good). Maybe they’ll subscribe an RSS reader to the feed and forget about it, maybe my UX is horrible and no-one will ever figure out how to change the users commenters they’re following. Either way, it’s a lightweight solution just in case it suddenly gets an influx of marginally attached users.

My default plan of action is now:

• Step 1: Make it work
• Step 2: Make it scale assuming users log in and use it once with the default settings
• Steps 3-N: ?
• Step N+1: Make it scale for real

In general don’t start people off with a blank slate, have a good default experience. Here it helps me cache more efficiently, but that’s a side affect of configuring it well for a new user.

You don’t get a lot of control over when (or if) you become popular, but the one thing you can count on is the fickleness of social media visitors. No matter how many visit on your big day (whenever it is) they’ll mostly just poke around, click once or twice and move on — it’s their second visit you’re interested in, and that’ll be a small percentage and more spread out.