Hacker News Reader

> For example, if the #1 person believes something is being said that is really damaging in some way, they could put at stake the entirety of their karma and downvote it, which would pretty much obliterate the comment.

We are building a P2P network with subjective moderation [0], so while we don’t know what works yet (we’re experimenting) I can at least tell you why we’ve found this is suboptimal. The issue is that this creates an even more direct USD/upvote market in which users are incentivised to the max to harvest as much karma as possible (with cat posts) and then sell them to the highest bidder, which would invariably be people who have a lot to gain from a certain piece of information being hidden. In essence, if you make karma spendable by more than one per post, you up it’s monetary value 100x, and given that even at 100x it’s worth so much more to PR teams and to not many else, all you will ever see is going to be propaganda on your platform’s front page.

[0] https://getaether.net

You've probably thought of this, but if you can associate votes to identities then a classic way to mitigate this behaviour is quadratic weighting, where e.g. your nth vote costs n^2, so a censor or propagandist needs to pay 1e4 counteract the single votes of 100 casual observers.

Yeah, quadratic voting is one of the potential solutions to this, but ultimately that removes the ‘fraud load’ by mostly moving it to another place, to number of accounts. That is arguably worse because it incentivises the malicious users to create as many accounts as there are votes so as to not be subject to quadratic voting (1 vote = 1 energy point equivalency only holding at one vote), and it’s substantially harder to deal with user account fraud than vote fraud.

The goal is to minimise the fraud load and make whatever fraud load you have applied against the strongest parts of your system. It’s like designing a castle except with incentives - it has to fail gracefully and still be functionally successful even in failure.

Adding other factors into this system might help.

What about:

    weighting a vote by the "degree of separation" of the voter? (ie., friends count more)

     hyperbolic discounting of karma acquisition 

     limiting the number of voters per item

    new accounts can't vote until achived a network & karam status (eg., shares, by friends, are upvoted). 

    accounts which primarily upvote/downvote outside of their network are ignored

    etc.

I think this adds up to saying, "small networks of voters" are the primary type of community; and big-network effects are largely discounted. Where things are popular, they are because many different active communities appreciate them.

Yes, these are all different kinds of logic that can be executed, and none of them are ‘right’, it just depends on personal preference and situation.

We call that user-selectability ‘subjective moderation’, but it’s basically a client-side compiler for a people-based raw data feed.

> ultimately that removes the ‘fraud load’ by mostly moving it to another place, to number of accounts. That is arguably worse because it incentivises the malicious users to create as many account

Arguably the common social media policy of not having your account tied to a unique identity needs to change. It permits all sorts of problematic incentives like you describe, and which lead to bot accounts and more.

Of course, your identity doesn't need to be disclosed in your profile for when anonymity is important.

> I think up and down voting are great

It is great as a poor man's analytics system, providing knowledge that at least someone read your comment, allowing you to evaluate if it is worth posting more or if your efforts are disappearing into the void. However, having both up and down seems redundant. The both provide the exact same information.

> however what is dumb is that neither has a cost.

If there is a cost then there will be reduced incentive to let you know that your comment was read, which reduces future engagement, which is not ideal for the platform.

> It is great as a poor man's analytics system, providing knowledge that at least someone read your comment, allowing you to evaluate if it is worth posting more or if your efforts are disappearing into the void. However, having both up and down seems redundant. The both provide the exact same information.

I disagree. If a comment has 100 likes and 100 dislikes that clearly is different than a comment that has 200 likes or 200 dislikes. A new metric, controversiality is created when you look at both the total amount of likes and dislikes and the ratio between them given a comment. Depending on your inclination a comment with a high, equal number of likes and dislikes is perhaps more engaging than the same number of solely likes or dislikes.

> If there is a cost then there will be reduced incentive to let you know that your comment was read, which reduces future engagement, which is not ideal for the platform.

Depending on how it's implemented, this may or may not matter. For example you could allow people to like or dislike without limit, but after a certain amount (them exhausting their own karma) its effect changes, but regardless they can continue to do it.

This is basically what people did on their own when they say a tweet gets "ratioed" by having more quote retweets than likes - the downvote button is probably just turning that into a first-party feature.

> I disagree. If a comment has 100 likes and 100 dislikes that clearly is different than a comment that has 200 likes or 200 dislikes.

There is no information provided by the button click other than that the button was clicked. Which button the user decides to choose may as well be considered arbitrary because onlookers have no way of ever determining the true intent unless the button presser also comments with their intent.

I find it curious that what was up/down has changed to like/dislike in your comment. Being able to see that someone read your comment is the reward provided by these 'karma' systems. Nobody wants to spend a slice of their life writing something that nobody will ever read. Finding readership is the goal and creating content with enough quality that someone wants to go out of their way to tell you that they read it is how you validate that you are accomplishing your goal. Either button can convey that information. With that, isn't 'dislike' logically no button press? If you truly dislike something, why do you want to reward the author?

I think only a troll commenter would find receiving dislikes to be a rewarding experience.

I imagine receiving dislikes wouldn't be rewarding, however the up/down buttons do not convey that kind of information. All they can fundamentally tell you is that someone (assuming appropriate bot protection is in place) pressed a button in proximity to your comment. There is no other information attached with the action. It could be purely accidental for all you know. Knowing that people have loaded your comment in order for the button to be presented, and thus statistically are likely to have read it, should be rewarding, though. After all, wanting others to read your work is why you're writing in a public forum and not your private journal.

Looks like you are being rewarded quite handsomely.

Not only is "there is no info provided by which button was clicked" a really odd take when we humans explicitly assign semantic to up/down vote, this contradicts your other point, which is that "having your post read is the real reward," since there is no guarantee the post was read before voting.

> when we humans explicitly assign semantic to up/down vote

What kind of semantic would that be? Using a real-world example, some of the comments I have made in this thread have had the up button pressed multiple times. Others the down button multiple times. There is no difference in the comments. The are all about the same subject, all written by me about that subject. There is no discernible explanation for the difference. All I know is that the buttons were pressed. There is no more information provided to me.

> this contradicts your other point, which is that "having your post read is the real reward," since there is no guarantee the post was read before voting.

How so? My "other point" (it is the same point) explicitly calls out that those guarantees are not made, using an accidental press as an example.

This comment being down-voted, along with the top most up-voted child being a quip about being rewarded for being down-voted, is art.

It's very clear that absolute vote count is a meaningless number. The vote count depends on which side of the argument that the reader belongs to. The simpler side of the argument will therefore attract more upvotes.

What the writer of the comment ultimately cares about is how many people were against the writer's comment that is now for it. Or at the very least, a piece of the argument has pushed the Overton window for that individual so that if they continue reading that writer's post and those whom allegiance with them, they will flip. That must be a vanishingly small number of the votes. Votes don't give you that.

It's very clear from my perspective that randomdata's post was made in good faith. randomdata's argument is one of clarity from understanding this nuance. He doesn't go into this depth, but it's there. The only way that you make the argument that randomdata is making is after having comes to terms with the mechanics. As he has skipped showing this depth his comment is unable to bring those of which understand the simpler argument but don't understand his argument, and he is therefore down-voted for it.

The fact that it has taken me five paragraphs to explain this nuance and not even go into it myself is why Twitter can be such a nasty place, as you don't have five paragraphs.

I'm now making this post 12 hours later, right when the post is at the bottom of the frontpage is therefore about to fall off a cliff of readability. As I care about getting my message across, I'm in complete agreeance with randomdata's argument. (+200 / -200) is much more important to me than ( +3 / -4). It's very likely the only person who might read this at this point is randomdata, of which I am speaking to the choir and achieving no change of view.

Note sure what you mean by "a new metric".

Sorting by "controversy" has been an option on Reddit for a long time. Most people won't use this option, but it exists.

The formula is here: https://github.com/reddit-archive/reddit/blob/753b17407e9a9d...

> you could allow people to like or dislike without limit, but after a certain amount (them exhausting their own karma) its effect changes, but regardless they can continue to do it

Reddit already applied this at the aggregate scale: any individual comment can only cost you up to 100 karma and negative karma scores on user profiles are capped at -100.

That all said, how would these karma economy proposals interact with the fact that some subreddits are orders of magnitude less popular than others? Would it cost less to vote on items in small subreddits?

Would upvoting already highly-upvoted submissions be free (but do nothing on the back end) or would it be expensive as an encouragement that the community has already spoken and to spend your votes elsewhere? Would doing against community opinion be cheap (since more people agreeing the Earth is round doesn’t add much to the discussion) or would it be a pricey 10x vote?

I like your ideas that voting should have a cost. I think though that your last paragraph (staking much of your karma) would inevitably lead to a karma market and marketers abusing the feature.

Then I guess the only thing we can do is create a Karma Exchange Commission to regulate karma transactions. Obviously we must provide karma wages to those working for the Commission, so we’ll also need to collect a karma tax for funding purposes.

I really do think innovative in social media spaces in lacking, probably because the established players have no real reason to change things up aside from stealing newcomers's ideas

but it would to cool to see if any research could be done on stuff like seeing how different features (or lack of) lead to different communities and cultures on a site

How do you define success for these experiments? Social media already is optimizing for engagement, probably the average hn commenter doesn't like those results though.

Agreed. Wondering why social media companies aren’t trying to make the platforms less engaging is like wondering why drug dealers aren’t making a less addictive opiate.

obviously every platform wants engagement, they want you to use their product over others! but I think that's only part of how you define success, which obviously will differ platform to platform

like I think something like "sustained high quality discussion" is good goal other spaces like HackerNews aim for

Matrix and the power of decentralized federated social networks will allow this soon.

Yes it will! https://github.com/KombuchaPrivacy/circles-ios

Does Matrix support full extensibility of data structures or are you tied to some minimal model?

full extensibility

It's also that they have reason not to change up things. They have created vibrant monetisation mechanisms & fine tuned those mechanisms.

Innovating means new patterns of interactions & ideas. This also means new forms of monetisation that might be required.

As an example all the previous FB & instagram ads became obsolete or useless once they introduced stories. Now, have to create better ad formats and subsequently take few quarters to fine tune the experience and pricing.

Twetch is an experiment in exactly this. Pay to like, comment, "branch" as they call it (retweet). pow.market is another fascinating experiment using proof of work as the currency that could be integrated into these platforms. ctzn.network is another take but no payments going on there. "Karma as a token" is what I think of based on your description and it's really cool. I think in the next couple of years there will be a lot more experimentation in this area too. It's a perfect storm of censorship, bots, and fake news that are encouraging more experiments now. There are others I'm aware of that are not yet public that I'm most excited about though. It's definitely happening.

I think it's important to acknowledge that in social networks attempting to manage their communities through mechanisms like this, there's a tacit assumption that everyone participating is

a: human and discrete

b: in good faith while of differing opinion still

This is not true. As long as social networks have a worldwide reach and a real-world significance, it'll be valuable to accumulate bots/minions to compose an aggregate person, representing some force or entity, and then use that brigade to direct discourse.

If it becomes expensive to do this, it takes the behavior out of the hands of small groups and reserves it for the most powerful and well-resourced actors.

The question to ask here is how expensive karma farming will be, and how easily can it be coordinated by outside forces in the absence of 'good faith, discrete, human actors'.

Quora used to have a cool "credits" system where you got credits for creating popular content and then you could use those credits to boost other content. So if there was an answer you liked you could spend your credits to expose that answer to more people. It was fun if a bit pointless.

> It's a shame there's not more experimentation here.

It is, but partly there's a good reason: it is really hard creating a system that is better, similarly convenient but also not more prone to abuse by bots and similar attempts to gain an advantage.

The more valuable karma gets for voting, the more worthwhile efforts to "farm" karma become. Ideally you want to maximise one thing and minimise the other but they are linked.

It's like with the recent case where a large-scale mining operation was discovered, but instead of cryptocurrency they used game consoles to farm currency for EA's FIFA game which can then be sold for real money because thanks to microtransactions and definitely-not-gambling-boxes it gained real value.

Stack Overflow did it quite well imho

there is no “group thinking” issue on StackOverflow. Quite contrary :)

I use voting to keep track of whether I’ve watched something before or not. There’s not really a good way to tell most sites you don’t want to see the same thing twice.

How about meta-moderation, like what is used on Slashdot?

Logged-in users are occasionally asked to apply a few moderation points to comments, but are also subject to their moderation actions being reviewed in turn.

>You can then give people more karma for successfully being a user without being flagged or receiving certain amounts of negative karma for a time period.

So a few users only have to band together and target one individual to limit their karma spend?

Sure but the entire point was that they’d have to pay for that privilege. Presumably you could make subsequent blocks exponentially expensive.

The source of karma is the attention people pay. That's why it's referred to as the attention economy.

Each user gets 10 votes per day, and can only vote up or down on a topic once?

Each user gets 10 points per day, and can use it in a single vote or spread across hundreds?