I first created an account on reddit about 6 years ago, and I think I had been a lurker for a while before that. When I joined, reddit was a place of learning. There was (to me, at least) a sense of community around sharing great philosophical articles, important news stories, and all kinds of learning resources. In essence, it felt like Hacker News before Hacker News. This probably isn’t a coincidence , as Paul Graham had a huge influence on reddit (which wasn’t even the idea Alexis Ohanian and Steve Huffman originally pitched for YCombinator). And over time, it lost that high quality aspect. I suspect that most people reading this post have heard this argument somewhere before .
I loved reddit for a long time. And now I love Hacker News. But there is something common between the two, and many other sites, that I think naturally leads to a decline in quality over time.
There is a quote on the guidelines page of Hacker News that I find amusing :
If your account is less than a year old, please don’t submit comments saying that HN is turning into Reddit. (It’s a common semi-noob illusion.)
It seems this must be a pretty common assertion for it to warrant a spot on this page. For the record, I’ve definitely read Hacker News for longer than a year, even if my account isn’t that old. I’m not here to slam HN, but comparisons between it and reddit are inevitable in talking about this.
Data, you say? #
I’ve created a little script to crawl cached versions of reddit’s front page on archive.org from random days, and compile some stats. It’s not pretty, but it’s pretty effective. My posting it is mostly to confirm that I’m not just feeding you bullshit, but I’ve made very little effort to make it particularly sophisticated or easy to read.
I decided to see how many posts on the front page were videos, images, internal (AskReddit, AMA, etc.), and “other”. In my estimation, this “other” category contains most of the content I was referencing at the top of this post. Well, this probably won’t surprise you, but “other” as a percentage of the total number of posts has gone down over time. How much?
By a bit, I’d say. I was surprised to find that videos were such a small part of the front page, but I was not at all surprised to see the rise of the image. In fact, in 2011 there were more images on the front page than any other kind of content. This “clicked” with what I already believed to be true, but I didn’t want to let my biases get in the way, so I ran these tests two more times with remarkably similar results. It’s not even worth including another graph - you will get basically the same graph over and over again if you run the script yourself.
I don’t want to dive down the rabbit hole of blaming the founders, imgur, or Conde Nast. I don’t think they’re actually to blame, because it’s the community itself that reaffirms it every day. I also won’t bother doing the same analysis to Hacker News (though it wouldn’t be hard to tweak the script to do it), firstly because I am ultimately happy with the content I get there, and secondly, the classification of “good” vs. “bad” content gets a lot more nuanced since images and videos are a relatively tiny part of the front page of Hacker News.
Caveats #
This data is not the most scientific in nature, and I welcome opinions about my methods. I’ve run my crawler many times and gotten essentially the same results every time. This should suggest that the year-on-year trends are pretty accurate, as the script was choosing (in my case) 40 frontpage captures per year, with multiple executions accumulating more links each time. I also started this process from scratch multiple times, with the same result.
I have used the front page of reddit for these metrics, because I feel like it is the most popular representation of reddit consumed by most, though you could argue that disregards all the great stuff happening in subreddits. In my view, subreddits are a symptom of this problem. People want to create their own little niche to escape the torrent of shit being unleashed on the front page. Thus, bigger subreddit = worse subreddit. Same problem.
So why does direct democracy inevitably suck? #
I’m anticipating a little skepticism here. I don’t consider myself above Paul Graham, the reddit guys, or anyone else doing their thing. I just think there is room for another experiment, and communities everywhere could potentially benefit from it.
I believe (and I’m not alone ) that over time, as the number of people in a community increases, the things that they’re all able to agree on become more and more rudimentary. If you take a community of 50 programmers, they’re probably able to agree on a lot of different topics that many of them find interesting in the fields of computer science, math, etc. Let’s add another 1000 programmers. There are probably more factions now, debating Ruby vs. Python, and rehashing many other classic programmers’ disagreements, and finding less and less that they can just agree on.
Now, let’s say for some reason that non-programmers begin to find the site interesting because there is a lot of ground-breaking research posted there, and it becomes the place where hip new stuff debuts on the internet. Now you’re really reaching to find common ground with the entire community. Let this process run for a few years, and eventually the only things people are able to agree on are LOLCATs. Okay, maybe not, but it’s clear that you have to do some cherry-picking to keep it from devolving this way. And this is why I think Hacker News has mostly weathered the storm without alienating its entire original audience: moderation. They also do things like intentionally keeping their design very plain, in part to help hamper growth.
What Hacker News is doing right now is ultimately unsustainable. Eventually, they will need more moderators, and those moderators might not always act in the interest of the entire community (because how the hell can they when they community gets to a certain size?). Thankfully, I think most people reading Hacker News would take a bullet for Paul Graham, so the idea of him or Sam Altman declining in respectability is unlikely. I do, however, think a decline in their availability is inevitable.
Another chuckle nugget from the Hacker News guidelines:
Resist complaining about being downmodded. It never does any good, and it makes boring reading.
So reddit/HN = direct democracy? #
Yes. The central idea of direct democracy is that everyone gets an equal vote on everything. This is true on reddit, Hacker News, and many other similar sites. Unlike most real democracies (total or partial), there is usually no minimum requirement to vote. The idea that this might be a bad thing isn’t really new:
Dictatorship naturally arises out of democracy, and the most aggravated form of tyranny and slavery out of the most extreme liberty. - PlatoA democracy is nothing more than mob rule, where fifty-one percent of the people may take away the rights of the other forty-nine. - Thomas Jefferson
In the United States, we decided early on to have a representative democracy, rather than a direct democracy.
Well, that’s great, until the representatives no longer represent the views of their constituents and have no inclination to do so. I would argue that the incumbency rate in the U.S. Congress proves that, despite the unhappiness of the people they represent, representatives in Congress have almost no reason to worry about being voted out of office.
This is why I think we need more fluidity in the moderation structure. And it’s why I want to launch my own experiment, that I think will create a more resilient and sustainable kind of news site. In my view, the only way to get the best of both parts (democracy & republicanism) is to have people choose as many representatives as they want, and have a way to “recall” the representatives if they no longer meet their quality expectations.
The ideal thing would be to let news creators, who are agreed upon by the community to be the best by the approval they get for each item they post, dictate the contents of the front page. I think these people should get more power than people with no respected content to their name.
The older internet dwellers among you may already be thinking “BUT DIGG SUPERUSERS WERE THE DEATH OF DIGG!” That’s true. Digg superusers had the same problem as ineffective representatives in the U.S. Congress. They are hard to replace, even when they are unpopular, AND maintain some degree of sanity. If you set the number at 10 moderators, or 100 super users, or what ever, you set the upper and lower limit at that arbitrary number, when it actually may vary wildly over time. And you’re also not letting the community really decide. It’s hard to constantly keep a group of 10 moderators on staff that never fail the community. You’re forced to base future expectations on past results. Basically, you’re both manipulating the currency and investing poorly.
Better, in my opinion, would be to fluidly let users become bigger forces over time, with that status being removed if they began to fail the community. Ideally, no one would have “staff” status - the community would be 100% self-policing based on the “currency” of reputation, with the ability to revoke that reputation if it was no longer warranted. Also, no one would be “fired.” User influence would just ebb and flow with time.
So HOW do we get there? #
I think it’s quite simple, actually. Votes are limited. You only get a certain number before you can’t take any actions anymore. You spend your votes like currency.
What if you left the selection of these superusers entirely up to the community, not by formal process but by constant re-affirmation through positive votes, and gave the power to them to take that status away at any time with negative votes? And what if you could give as many of those votes as you wanted at a time?
The content creators from the beginning would gain the most votes simply by having the longest time period to have contributed in. This would skew the overall “wealth” distribution towards early adopters (the group of people whom everyone else ultimately seeks to join, but can’t all be a part of by definition). But, that wealth can’t be socked away and hoarded to the detriment of everyone else - at some point, people have to be able to “tax” them in some way if they aren’t contributing in a constructive way.
Basically, I’m talking about a new capitalism for content.
I want to explore what this could look like. It could even go so far as having cryptographically signed votes that can’t be manipulated by the staff of the site and can be publicly audited like Bitcoin. I don’t know. I have several other ideas for controlling overall quality that I plan to discuss in a future post soon. If you’re interested in joining that conversation, find me on IRC or Twitter .