Starting in 2017, Facebook treated emoji reactions as more valuable than likes | Hacker News Reader

Starting in 2017, Facebook treated emoji reactions as more valuable than likes

washingtonpost.com

65 points740222 comments

I know this will be downvoted, but I think it's worth noting a few points:

- All "reactions" were weighted higher than "likes", and the headline seems to mislead readers into thinking that only the angry-reaction was given extra weight, which is not true

- The data is from Facebook's own study, indicating they want to understand (and perhaps improve) user experience

- The study was from 2019, Facebook might have changed since then (my own newsfeed is overwhelmingly positive and not sensational so it could be)

I think it's justified for reactions to be weighted more than likes. To put a reaction requires comparatively 2.5x the effort as a like (hit like, choose the reaction, select the reaction). By what factor it should be increased may be a matter of differing opinion.

Exactly, this is such a well duh, that I'd be surprised if it weren't this way.

On the first point, I agree, the article was very much misleading and I hope they correct it.

As for the rest, frankly, this just adds more fuel to my belief that the first thing we need from these companies is simply transparency. If they're running mass psychology experiments on people, we should know about it.

The idea that Facebook can operate as a black box while literally weighing how much to value individual human emotions is deeply deeply disturbing me.

Yeah, especially for a company that people spend hours using and that influences electiobs

In my opinion this makes sense? To react with an emoji, it requires more interaction from the user (tap and hold instead of a simple tap), which you could make an assumption that it signals it received more of the user's attention and is more attention-worthy content. This can be both bad and good, but is not inherently bad in itself. I can see why some would argue that an angry emoji could lead to a more negative news feed, but five out of the seven emojis you can react with are "positive", so you could also argue that positive content would drown out the negative content (assuming that there's an even distribution of interactions between emojis). At the end of the day, this was part of a study to understand human behavior, and without the study they would have no understanding of how emoji reactions could be used to improve a user's feed and improve their platform.

The irony of WaPo's coverage of this topic using a misleading baity headline to sell outrage clicks ...

Our legacy media corporations ought to dedicate a week of heavy coverage to a digital ad industry whistleblower and encourage everyone to install uBlock origin/Pi-hole for their digital hygiene. "It's like wearing a mask online." This would do a lot to change the incentives that encourage crap viral content.

I'll hold my breath.

http://web.archive.org/web/20211026133005/https://www.washin...

It's not clear, did FB steal that formula from traditional media, or was it the other way around? Or was the discovery of the value of extreme simultaneous? Traditional media loves hyperbole, half-truths, etc. Whatever it takes to drive revenue.

FB and its ilk are not benevolent public services. They maximizes engagement and in turn drives ad revenue. Full stop. Let's not be naive. There should be no surprises about their biz model and drive.

That's not giving Big Tech a free pass. But we all know - or should - these services are not our friend. They have a purpose and that's to maximize shareholder value.

And of course, Jeff would prefer the dirty spotlight to be on Mark. So let's not ignore the source either. The pandemic has served Jeff well. Let's not talk about that.

The article states:

>"The average post got a score of a few hundred, according to the documents. But in 2019, a Facebook data scientist discovered there was no limit to how high the ranking scores could go."

Might someone be able to say how the fact that scores could be unlimited was only discovery well after this system was implemented? Wouldn't the upper and lower bounds for scoring be part of the algorithm design?

Yes, we know FB is bad. This is preaching to the choir.

I want to read an article about how to fix things.

How exactly could someone propose a fix without knowing any insight into the root cause of the issues

I have no doubts that Facebook is bad, but I do think it's worth pointing out that a lot of these articles are being written by Facebook's competitors. Washington Post certainly has cause for some ulterior motives on the topic.

I see two actions being talked about more and more (including by Al Franken on TV this morning).

1. Anti-trust action. Split off Instagram, WhatsApp, Oculus, and other companies that they acquired.

2. Section 230 modifications making them culpable for decisions made by their algorithm.

A possible fix.

If someone owns more than 40% of the market in any service area / aspect, regulate as a utility.

Force open documentation of processes and practices (including weights for algorithms, etc).

Force interoperability requirements to break lock-in and make it easier to use other tools and services. (Anti AT&T device rental / must use the closed source client software.)

If above 60% market share in a service area limit profits as a price-regulated public utility.

What would this mean for a small town? No one would want to open the first bakery, the first hair salon, the first bait shop, etc.

Insufficient population to establish independent relevance.

Probably the 'market' would be a county or set of counties in that case. Likely within that region the only monopolies (in the US anyway) would be Walmart, chains of stores like 'dollar general', and maybe tractor / farm supply services.

Yea, this time is different. It's just about selling newspapers.

Cambridge Analytica was a scandal that boiled down to: FB gave your data to other people without asking you. The EU, state of CA, other states passed privacy laws that companies will get in trouble for misusing data. Newspapers made a bunch of money, meaningful change happened.

In this case, what's the core issue? FB sometimes makes people feel bad? Spreads misinformation? How are we going to fix that? Fox News spreads misinfo. Cosmo magazine makes people feel bad about their bodies. Are we going to start running sentiment surveys for every news or entertainment product so we can shut down the ones that are above X%?

Common theme of the articles coming out now just seems to be "We got some inside dirt, it's going to make you angry, come take a look." Gonna sell some papers, but I don't see what the meaningful change will be.

I know some want to shut down or break up FB, maybe that's the endgame of all this. I don't see a world where driving kids further onto Snap, TikTok, Reddit, Discord, whatever is gonna change anything about misinfo or how they feel.

The solution is pretty simple - get Zuck and his mindlessly ambitious crew of FB execs trapped in the metaverse as soon as possible.

A basic program on loop, can tell them every few seconds how brilliant, good and universally loved they are, how their net worth is growing faster than the big bang generates matter, and how without what comes out of their 6 inch chimp brains the universe itself might stop existing. That hopefully is enough to satisfy their needs.

Actual article headline: "Five points for anger, one for a ‘like’: How Facebook’s formula fostered rage and misinformation"

Though I'm highly sceptical of FB, title is misleading clickbait itself. Should be: 'five points for emoji, one for a like'.

IMO that's an even worse rating system.

'drunk hieroglyphics' should be at best equal weight to reacting at all with an up ( or down, but I disagree about that being useful ) vote. If such vote buttons exist, flag for moderation should also be a button, maybe something symbolic for calling the cops (flashing light icon?).

Maximum points should be awarded for long, well phrased replies in discussion / engagement with the content.

To better society, the goal must be to encourage critical thinking.

Reply on news.ycombinator.com