I’ve seen this opinion, especially in the post-2016 world. I’m not sure I agree completely. I think it totally misses the core issue, which is that somebody was incentivized to be paid to do this work. If you’re an engineer at Facebook who thinks the News Feed algorithm is tearing people apart, quitting is not going to solve the problem. Facebook has an incentive to make the algorithm the way it is. Your refusal just leads to your replacement with someone who will do the work (and get paid well for it).
That’s not to say that you don’t have a moral responsibility to stop doing work that makes the world worse. But “all software engineers in the world should refuse to work on bad things” is hopelessly naive. If you want Facebook to stop showing people crappy news, find a way to stop Facebook from getting rewarded from it. Write laws. Raise awareness. Write a competitor that does better. But refrains like these, I worry, are just hoping that everyone will commit to doing the right thing 100% of the time. There are better solutions.
It turns out that using ML to optimize for immediate engagement has two unintended side-effects: 1) it produces junkier content, 2) it decreases long-term retention. For obvious reasons, building a model to optimize for the long-term engagement is way harder and takes way more time.
While in the long run new model is more profitable (due to increased retention life-long engagement goes up), it decreases immediate engagement metrics. When this happens, major accounts start to call in and ask why now they are getting less for their dollar, thus this preemptive explanation by Mr Zuckerberg.
Therefore, perhaps there should be a revision to popular open-source licenses that essentially says “not permitted for use in developing X” where X is “a social-network product” or whatever. This forces (law-abiding) companies to spend their own resources in more situations, increasing their costs.
1. Joel Spolsky, co-founder of stack-overflow, and well-known software commentator disavows twitter/facebook.
2. Theorizes that twitter's low-character limit leads to misunderstanding, which leads to conflict, which leads to drama, which is addictive to our "rubber-necker" tendencies. Cites personal experience and hashtag wars in a comedic example.
3. Furthers that facebook's machine-learning is (advertantly?) accomplishing the same effect, by selecting for things that elicit a strong response (mistaken as "engagement").
4. Proposes that facebook uses "intermittent reinforcement" to ration out its good content slowly, forcing users to spend excess time on the site to get to the small amounts of interesting material they enjoy.
5. Proposes that these naive attention-optimizing algorithms can lead to much "shriller" political battles. Raises a moral obligation to engineers to acknowledge the large-scale social consequences of simple algorithms.
Joel criticizes twitter's forced summarization because it leads to oversimplification and resentment. I don't think he's arguing that all forms of summary lead to oversimplification and resentment.
Think of it as bullet-point abstract to journal article.
And by the by, any time you find yourself using words like "all", "always" or "never" in describing someone else's position is a good time to pause and reflect on whether you're honoring the Principle of Charity, as paraphrased in the HN comment guidelines thusly:
> Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize.
The good faith interpretation of my post was that I was complaining about your summary, not about summaries generally and certainly not about summaries in wildly different contexts like journal abstracts.
However, just to reiterate: please understand that I'm not attacking you, or saying that you mis-characterized anyone's position on purpose, or out of malice or incompetence. The whole point of the essay we're discussing is that conveying a complex and nuanced position through a short block of text is hard, and that when someone tries, a distorted view of the author's position is the expected outcome.
For a while, I tried the "Show me less like this" response. I slowly came to realize that this button doesn't do a damn thing (actually, it's worse). Facebook didn't stop showing me this kind of news, at all. At the time, I figured "show me less" was like those crosswalk buttons that don't have any actual influence on the timing of lights, but deter jaywalking by providing an illusion of control to a pedestrian. I realize now that it's probably worse than that - "show me less like this" is probably the Facebook equivalent of asking a spammer to take you off the email list. All it does is tell them they have a live one.
I am glad to see someone definitely smarter than me find this opinion too. As far as I can tell the only value in twitter is network effects and the total removal of nuance.
I think this is why I continue reading a few dozen webcomics every day. Some of them are good, but oh boy is that caffeine kicking in.
I found that the things that were making me angry were the reposts of garbage by friends and family who don't apparently have an original thought. If someone posts a cogent comment that I disagree with, I'll be happy to discuss. I don't see that very often. I just blocked viewing for the people that posted that kind of crap. I do miss the occasional interesting thought that they post but I also don't see all of the garbage.
Yes, this puts me in the bubble of blocking the things I don't agree with, but very little of the stuff I disagree with is anything beyond an inflammatory repost.
Ok, Joel obviously isn’t talking about puppies. What is he actually referring to? Anti-immigration comments? Casual suggestions the wage gap is a myth?
The overt message he’s giving is “I am scared I might say the wrong thing and get hurt.” But the subtext is: “I think the people currently being removed from their jobs deserve to stay in them.”
But he doesn’t have the guts to say which people deserve their jobs.
The result of that is he signals support for all of them, from Eich to Weinstein.
I feel like we need to be more surgical than that in 2018. Not just throw our voices in with movements that are too big to be bound by any kind of principle.
I'm guessing this:
Although Spolsky didn’t call it out, I agree with you that his sentiment is likely that she shouldn’t have been fired. And that’s almost my opinion too.
However, to conflate that to meaning support for Weinstein is a rather large leap and one that I don’t think is constructive to the discussion unless there is more evidence. As not it’s largely speculation, so it’s impossible to resolve if we go down that path.
This goes toward Spolsky’s point that social media (partially HN as well) leads to more shrill discussion. Perhaps ubconsciously so, I see things commonly conflated with their most extreme possibilities and then arguing against this extreme straw man. For example, “you don’t think the hiv comment lady should have been fired? How can you support Weinstein? How can you think nazism isn’t bad for America. Your silence is part of the problem.” And then tens of thousands of characters pro and against. While most sites then rake in ad revenue from the rage.
It’s hard to think of ways to change your own behavior. For me, it’s avoiding commercial social media sites and favoring simpler sites like HN.