Hacker News new | ask | show | jobs
by cortesoft 2404 days ago
One of the reasons that reddit can avoid 'blog spam' is because as a centralized service, they can control user registration and take measures to prevent bots from signing up for thousands of accounts and upvoting their spam.

How would this be prevented in a decentralized system? Spammers would join the network, create thousands of bots, and upvote all their spam.

4 comments

Reddit doesn't really avoid anything as far as I can tell. A lot of the content in the big subreddits is more or less well concealed "organic advertising".

The upvote system is also basically useless on reddit these days, the site is way too big to enforce any kind of etiquette and people gladly upvote empty one-line quips while they'll bury anybody expressing any form of vaguely dissenting opinion. It's frankly toxic.

At least good old blog spam is easy to identify and ignore.

There are a few topics I care about where you have both big subreddits and also dedicated "old school" forums on third-party websites. The forums are always more interesting and in-depth in my experience, when the subreddits are just a litany of karma-whoring picture posts and low effort so-called "memes".

One of the things I prefer about HN is people seem to use the downvote button more often as a "this is a poor quality contribution" vote, rather than a "my opinion differs from yours" vote.
That hasn't been my experience. People show their dissent of opinions with downvotes here on HN too on sufficiently polarizing threads. I think the main difference is the type of content posted on HN vs reddit and the type of people who readily consume that content.

For example, I have noticed shameless downvoting in operating system war threads, both on reddit and on HN. In highly technical threads about physics or mathematics I have noticed people tend to vote based on the quality of the content posted.

Also, you need a lot of rep to get the right to use the downvote button, which also removes a large chunk of the would-be-downvoters

   The upvote system is also basically useless on reddit these
   days, the site is way too big to enforce any kind of 
   etiquette and people gladly upvote empty one-line quips 
   while they'll bury anybody expressing any form of vaguely 
   dissenting opinion. It's frankly toxic.
This.

I don't know what the rest of the people praising reddit's redesign are talking about. My experience has been that its so heavily sanitized that unless you say something saccharinely in-tune with the consensus view, your comment risks being banished to the nethermost regions of the thread.

Most of the comments are rehashes of the same groupthink, echoing the same banal and glib points over and over till they choke out any real thought provoking, refreshing and original insights.

If the cost of having a system that surfaces original and thought provoking viewpoints means living with _some_ borderline hate speech subreddits, I dont mind paying that cost.

The alternative is just feel-good insanity and a thousand times worse, in import.

I enjoy Reddit for small communities, with content matter expert moderators, etc. The default subs are effectively useless other than entertainment.
"Decentralized" is the key here. One option, already mentioned, is federation like Fediverse. Even more interesting would be friend-of-a-friend system, where rating of each post is subjective, and weight of someone's upvote is inversely proportional to degree of separation between you and them.
Instances can A. block bots on themselves, B. block other instances known to harbor bots the same way they ban bots on their own instances, and if necessary C. whitelist federation if there is a pandemic of hostile actors.
I wonder if crypto could play into this. Blockchain prevents sybil attacks by making interaction not free so that large scale influence is incredibly expensive.
This was tried: steemit.com. It didn't become very prolific, perhaps because the daemon was difficult to run, and there weren't many good frontends. The founding organization subsidized account creation, but was bad at spam detection, so eventually gave up.
I mean couldn't you just require a phone number for all accounts?
Last time I checked you didn't even need an email address. username & password only.