Tuesday, July 23, 2013

Filtered Facebook vs Searched Twitter

I've been ruminating on the differences these two social media in terms of what users can and do see.  This post starts by discussing this general comparison of filtered vs unfiltered media, touches on why  FB filters and then, focuses intensely on issues in the FB edgerank algorithm.  I'd appreciate detailed answers to the questions raised here. Thanks.

When people go to Facebook, they arrive at their personal news page which has updates from friends. And some advertising.   But, many people are surprised to learn that it does not have all the updates from all their friends. Facebook decides which ones you should see.  Why does FB do this and how do they decide?   Well, here is Facebook's problem, as people get more and more friends and their friends have more and more updates, the user experience deteriorates and the news and updates that you really care about end up getting lost.

Just think about it, how many FB friends do you have? I just checked mine (which turned out to be a harder than I thought: I found it by clicking on my name on the top right, then there's a tab below that has friends.) I have 332 "friends." If they each updated once a day and I check it once every other day, there would be a lot of post to wade through before I found what I'm mostly looking for. BTW, I'm mostly checking FB to see what humiliating picture of me my wife posted.

Facebook however has thought about this and has a system (an algorithm) for predicting what updates I would most like to see. They check on both my relationship with the person posting and generally, how popular that specific post by a person is. More specifically, they predict my interest in seeing a post based on three factors:
- which people do I tend to comment on, share, click on, or like?
- whether a post has been endorsed in these same sorts of ways by others?
- how recent is the post?

In contrast, Twitter seems to be an unfiltered feed.  Every tweet by everyone that I follow just streams through my feed.  This is the reason hashtags, a way of making searching easier, is such a big deal on Twitter.

I'm in business and I try to make sure that my company's Facebook posts on our company page, are seen by as many people as possible. So we pay attention to the possibilities.

We could (and sometimes do), just pay for more visibility.  FB helpfully suggests that I "promote" my posts by entering my credit card. They suggest this to me many times a day and several times a month, I do sometimes give them a $100 or so (not counting what our marketing department actually spends which is far more systematic and has more zeroes on it).  Small potatoes but for FB but it does learn me into starting to spend.  I often wonder if they could learn from Google who early on, lured me into spending $5K a month with them via the gift of a Google refrigerator.

I'm thinking it's time to upgrade our FB effort to increase visibility.  On the manipulative side, I wonder:
- How easy it is to manipulate this by having lots of people comment on and share a post immediately after posting.
- Does 2 immediate interactions matter or would it take 20? 200?  Do they want a natural pattern and what would that mean?
-  I wonder what "right after" the post means.  Do reactions in the first few seconds matter? First few minutes? What's the granularity on this?
-  I wonder if FB would notice if all the commenters were all in the same area or on the same IP address?
- I wonder if which interactions matter the most. Do we get the maximum exposure by commenting, sharing, or commenting and including a link?
- Does FB notice patterns and does that help or hurt? For instance, if they showed my previous two posts to 10K people (which is a lot), are they inclined to help or hurt my next post?  Is there no relationship?
- Once FB notices that I pay sometimes, do they penalize my unpromoted paids to get me to pay more?
- Do personal pages and company pages behave the same?
- Do endorsements by people, acting as themselves but also who can be FB page admins, count more or less?
- Do endorsements by all people on FB have the same impact? Is it weighted by the number of friends that a person has? Does each person have a limited "budget" of endorsements (sort of like the Google juice used in SEO by websites) which they can "spend"?
- How long does the initial burst of interest last? For instance, if we got 20 people to comment on a post in the first minute, does that momentum carry on even if the next 1000 do not comment? Would 5 in the first minute, and 5 more over the next three hours work better?
- Does FB track total interactions? Interactions per minute? Interactions per minute per person who viewed it?  For instance, if I post at 5am, there's not much going on on FB so it might be shown to all of our page's fans who happen to be on FB and refreshing their page between 5 and 5:15. Lets say that its 10 people.  If 5 of them comment, does FB think in terms of quantity 5 comments or do they think in terms of 50%?  Assume for the moment that FB just does a dumb count of 5 and says that it's a lot, so they show it until 7 am and another 500 people see it.  Will they keep promoting the post if just another 5 comment?  What, in FB's mind, is a high number or rate of interactivity?

More organically:
- what time of day do our users tend to be online and based on that, when should we post the most?
- does pointing towards resources on VSC get more reactions? Asking for their comments on a specific resource? Asking for them to suggest their own favorite resources? Asking them to comment on resources that they would like?

Mechanisms
- when we do our beta and other free give-aways, can we put some strings in place to get a cadre of teachers willing to share and comment on a regular basis?
- who is in the best position to drive and review this research in house to figure it out?


Toolchain:
- FB scheduler
 web.mention.net.
- hootsuite

No comments: