Hacker News Reader

Comment by burlesona | Hacker News Reader

> Google Brain’s researchers wondered if they could keep YouTube users engaged for longer by steering them into different parts of YouTube, rather than feeding their existing interests. And they began testing a new algorithm that incorporated a different type of A.I., called reinforcement learning.

> The new A.I., known as Reinforce, was a kind of long-term addiction machine. It was designed to maximize users’ engagement over time by predicting which recommendations would expand their tastes and get them to watch not just one more video but many more.

It saddens me that so many of our smartest engineers are working on stuff like this.

View on HN

> It saddens me that so many of our smartest engineers are working on stuff like this.

Does it sadden you if it's instead worded as

"The entertainment platform worked on improving their promoted content selection to prevent users from getting bored".

Really, we can have a discussion about ML algorithms running wild and achieving far different results than intended, but trying to exploit every angle for the sake of the "evil, greedy people" narrative feels very dishonest.

> "The entertainment platform worked on improving their promoted content selection to prevent users from getting bored".

This summary crucially omits that the platform's goal was not to prevent users from getting bored, it was to make more money. They were optimizing for time spent watching ads, not user experience.

I'll refrain here from passing judgment either way, but it's an important clarification to make.

I don't see whats wrong about suggesting videos to users that they like to watch.

Well, it catches people in thought-bubbles.

But the idea with this algorithm was to expand peoples interestes. That should work against locking people into bubbles.

"Get outside your bubble of rational reality and consider this crazy conspiracy nonsense..."

Same.

There is this pessimist view that Google is evil.

In my life, Google is the single best entity I'm aware of. Incredible amounts of free stuff that makes my life better.

Any addiction to their platform has been due to bliss.

Should have named yourself "totallydontworkatgoogle."

For one, it lowers their intellectual curiosity and makes them lazy consumers in a feedback loop of their current preferences...

Second, it always adds an algorithmic/editorial spin on the suggestion, so it's worse than them picking things up for themselves...

I see it from both sides. I love browsing youtube and finding relevant suggestions, I can kill hours of time clicking and watching. However I sometimes feel like a victim of some kind of manipulation. Youtube has incredible power to persuade you down certain paths (in the pursuit of maximizing # of views).

I think there is a difference here compared to other platforms: they slowly nudge you down paths of interest and you willingly oblige in watching it. And if you have lost interest in a subject, don't worry, it will pop up on your home page sooner or later, dangling it above your head "still interested? come over here, click on me!".

Sure I paint a sinister picture, perhaps it is exaggerated, but I truly believe that google knows more about me than everyone that I know personally, and maybe even myself. That is scary. If they combines all this information across their platforms, they are in possession of an incredibly accurate profile of myself. If they use this profile to suggest videos, I suppose it may not be as harmful as I have stated, but what if they were to use it for other purposes in the future? This is my greatest fear.

Those aren’t the world’s smartest engineers.

So now the idea is to make it so that the only algorithmic pathsways to information are to the sanctioned ideas? That's dangerous! Basically, in 2019, you don't have to burn books. The powers that be can bias the power of discovery and viral spread towards the ideas they like and away from the ideas they don't like. You don't have to suppress, just make the ratified ideas orders of magnitude louder. Or, that was the theory. Now they have to resort to mislabeling and mass deplatforming. This mental control scheme is failing just like the Soviet's control over official media failed: People will eventually see through the manipulation and stop believing it.

That's not to say that damage won't be done in the meantime. What's driving society crazy isn't some single percentage slice of economically disadvantaged under-educated fringers with theories just about everybody knows are crap. What's driving society crazy is the meddling in the free market of ideas. It's the breakdown of discourse itself. It's all of the perverse incentives to communication and other manipulation perpetrated by big tech firms that may or may not be monopolies. It's the meddling with people's speech by entities that aren't governments, but have more power and more tech savvy than most of them.

https://www.youtube.com/watch?v=rE3j_RHkqJc

These big media/technology companies have been amplifying outrage, amplifying outrageous behavior, bubbling people into echo chambers to further amplify the outrage, and aiding and abetting the toxic mislabeling of ordinary views and formerly protected forms of discourse -- which just adds even more amplification in the form of outrage.

The problem is not the single-percentage fringers without access to economic power, the halls of power, and the ear of those who control the newest forms of media technology. They are merely a symptom. Look instead to the politically skewed access. Remember: power corrupts!

Look at the magnitude of how crazy society's gotten in the past several years. Who has the wealth, algorithmic/technological power, and reach to do all of that? The answer's simple. Look for the tickers with the multi-billion dollar valuations!

All society needs to fix things, is to actually have Free Speech and a level, non-manipulated free marketplace of ideas.

Reply on news.ycombinator.com