As a blogger for (I think) about 18 years, and a BBS/Usenet user for almost as long before that, I have a lot of opinions about how people should interact online. To some extent those opinions are based on solid thinking about how the nature of that interaction shapes the discussion, and even shapes us. To some extent it's based only on personal preference and habit, which itself is often based on technical necessity in some no-longer-relevant context. Now I'm an employee at Facebook, whose product is very different than M-Net or Usenet 30+ years ago. Naturally I've been thinking about some of those differences, and one of the biggest in its impact is one of the smallest on the screen - the "like" button.
Before we dive in further, it's worth thinking about how Facebook differs from some of its predecessors. One thing that's not that different is the nature of the content being shared. I know this might shock some of the youngsters, but political discussions and joke threads and all the other things that happen on modern social media also happened in the early days. Many of the best lines from that era were memorialized in the "fortune" command that's still available on most UNIX-ish systems. Grab one, slap it on top of a random image, and it would fit right in with all of the recently produced memes. No, it's not the content that's different. It's the format, of course, but also the immediacy (latency to you geeks) and the reach and most of all the scale that's different. Scale matters. Optimal communication styles for two people, ten people, hundreds of people, tens of thousands of people, etc. are all different. I'll come back to that point later.
The main argument against "likes" is that they're too easy and convey too little information. Note first that those are two different things. Something can be easy and yet provide a lot of information, or it can be hard and yet provide very little. Let's consider the two separately - or actually let's not, because I don't think it's really worth discussing whether being easy is a good thing in this case. Let's focus on "too little information" instead.
Does a "like" really provide too little information? In what sense? It's clearly not too little to be worth transmitting and storing, because those costs have become trivial even in very large aggregates. The "don't waste bandwidth" argument from Usenet days is instead a waste of my time. Is a "like" too little information in terms of the cognitive load for the recipient to see and process it? Now we're nearer the mark, but to answer this question properly we have to think a bit more about both the costs and the benefits. First, what are the costs?
Seeing the "like" on the page: trivial. For one thing, you don't even have to look at that tiny little dot on the screen. For another, any number of likes are combined into a count, which is pretty easy to parse.
Getting notified: not trivial. A person who posts rarely, and only within a small circle, might not notice this cost. However, I can assure you that it becomes a real issue for people who post more often to a larger audience. Oddly, most of the negative sentiment seems to come from people who are in the first barely-affected group.
Now let's look at the benefits.
A "like" can make you feel good. A lot of "likes" can make you feel even better. This is real, but to be honest I think it's the least important benefit.
A "like" creates or sustains a sense of connection to a particular person. I think that's actually quite valuable. These connections do matter, and do atrophy over time if there's no reinforcement at all. If anyone even disputes that, I'll go look up some of the neuroscience behind it.
A "like" provides a positive signal to the recipient about how their content is being received. If all of my humor posts gets lots of likes and my political posts fall into deafening silence, how is that going to affect my behavior? Doesn't take a genius to figure that one out. Sure, the effect can be abused, but overall feedback and reinforcement are good things. (Side note: this mirrors many debates about anonymity, which can be beneficial but online tends to show a much darker side.)
A "like" also provides a positive signal to the system about what content I might like to see more of, what content is worth showing to my friends, and so on. Yes, that includes what ads to show me, but I think that leads into a whole other discussion so I'll just leave it there for now. As with the previous point, I think this also raises the issue of whether a "dislike" would also be a good idea (negative reinforcement has its value) but I'll leave that alone for now too.
Leaving aside the notification issue, I think that's a lot more benefit than cost. Notification is the only real issue, and that's where scale comes into play. (I told you I'd get back to that.) It's pretty well known that Facebook is one of its own biggest users. A lot of internal communication happens that way. Even though I'm still in boot camp, my feed is already more work than not. The notification light on my work-issued phone is practically always on. This is only going to get worse as I get further into my tenure, and it would be worse still if those "likes" were comments instead. That might seem like a solution to a low-volume user, but to a high-volume user it's worse than useless. A "haha that was funny" comment barely conveys more information than pressing the "like" button, and it's way worse in terms of the notification effects. Comments are harder to filter, aggregate, etc. If people commented instead of liking all the time, I'd hate you all. The light weight of "likes" vs. comments is exactly what makes them better most of the time. There's a time for likes, there's a time for comments, and there's a time for huge multi-paragraph posts (like this one). None should be used to displace the other. A comment that should have been a like is just as bad as a like that should have been a comment. Save the words for when they're needed, and don't be afraid to hit the "like" button more often.
Coda: the notification problem still needs to be addressed IMO. Delays and aggregation don't work, because at scale there's no delay both long enough to reduce the notification load and short enough to keep notifications useful. What I'd really like (heh) is notification filters. Notify me for each hundred likes. Notify me when I get a like from somebody who hasn't liked me in a long time, or from anybody with more than X followers (hello Mr. Zuckerberg), or according to some other criterion I haven't even thought of yet. Notify me for likes on this particular post. Otherwise, don't bother; I'll wait for the daily summary. Who knows? Maybe some of this already exists. Maybe someone's already working on other parts. Maybe some day I'll be involved in implementing other parts myself. That's one of the cool things about being on the inside ... but that's another post for another day.