A couple of days ago I wrote about what I perceived as a problem on App.net. Finding actual interesting people on App.net through recommendations. Over the next couple of days my timeline on App.net has exploded with people talking about this topic.

I was concerned about finding interesting people in App.net given that there appears to be a high number of bot and feed accounts recommended to you in the Recommended Friends section of the App.net Passport and on Alpha. I just want to find interesting people, not bots or feed accounts.

Having had a conversation with App.net's founder, Dalton Caldwell (@dalton), over the course of Tuesday and Wednesday on App.net, it is clear that this isn't so much a problem with App.net but a problem with all social networks.

I checked out the "People You May Know" block on my LinkedIn page. Initially it only showed me three people, but none of the people in that list were people I knew, worked with or even worked in the same company as them. It was only until I expanded the results further that I started to see people that I had recognised or work with previously. However, the results still showed that the majority of people there weren't people that I actually knew.

I remember the same widget also displayed on Twitter when I had an account there. Quite often, you could see that the people that Twitter recommended to you was based on your bio, your tweets and probably other information that Twitter graph internally. It wasn't completely accurate and often you wonder get a recommendation to which you would respond, why? Why is that person being recommended to me?

It's quite simple. Machines are not good at making relational decisions. My new friend on App.net, @novia, pretty much summed it up for me.

@matthewlang long ago my answer to a question about why computers can’t figure out simple things like this: computers suck at relational reasoning: Even if we create contextual awareness, it would take years of training & they’ll still make errors.
— App.net by @noviad https://alpha.app.net/noivad/post/12500139 App.net %}

So if machines suck at recommending other people for you to follow on any social network based on your own preferences, then how do you find the type of people that you want to follow?

The answer is simple.

Engage, take part, chat, converse, interact.

Be part of the social network rather than just a bystander.

Looking back at my Twitter days it took me a good couple of years to get my list of friends down to a list of mixed interests that suited me. Based on tweets, interests and location, I managed to get a nice timeline of like minded, interesting people to follow. I did this by tweeting to people, searching for hashtags in profiles and finding users or lived in my part of the world.

The same can be said for App.net, but there's a bonus with App.net. Conversations are indeed richer there, and that's where you find interesting people. By taking part. By contributing your ideas, opinions and thoughts to conversations. In order to find interesting people you have to seek them out. You have to intrude (in a nice way) in conversations and take part. It's through this action that I found a few more interesting people to follow last night.

Looking back, I'm actually glad that there isn't an all singing all dancing recommendation tool for people on App.net, because it would take the fun out of finding people on the service and connecting with others. I still think that there should search facilities for people on the service and that perhaps the recommendation feature in the App.net Passport have a filter that allows you to only see recommended people, but that would be all that others would need.

Thanks to Dalton and everyone that contributed to this post through their many conversations on App.net.