jwarden

11 ms·

91.
by jwarden 3y ago
This is my latest article about what we have been working on over at https://social-protocols.org We have had some good discussions on HN on some of other posts about improving ranking algorithms in social platforms. - Show HN:
92.
Paying for Truth: Information Elicitation Mechanisms(jonathanwarden.com)
1 points by jwarden 3y ago|0 comments
93.
by jwarden 3y ago
I'm sure it does, and I wonder how information about users, such as country, might be worked into the model as non-latent factors.
94.
by jwarden 3y ago
Okay I have added some examples to the post: http://jonathanwarden.com/understanding-bridge-based-ranking...
95.
by jwarden 3y ago
One possible explanation for why these dimensions don't improve the algorithm a lot is that differences in these dimensions don't cause differences in whether users rate a note as helpful or unhelpful -- at least not beyond what c
96.
by jwarden 3y ago
Yes, this is really an important point. In some online forums, the factor that explains variation among users votes may be precisely the thing we don't want to factor out. It may, for example, be expertise! I discuss this a bit in my f
97.
by jwarden 3y ago
Note that I am aware of. But me and some colleagues at social-protoclols.org have have written a number of articles on the subject. For example, here's an article on using Mechanism Design to solve the problem of dishonesty and promote
98.
by jwarden 3y ago
Yes I think this actually could work. Community Notes has a basic reputation system: users need to "Earn In" by rating notes as "Helpful" that are ultimately classified by the algorithm as helpful. Once enough attackers
99.
by jwarden 3y ago
Hi Aviv, thanks for replying and for the additional links. I think this work is so important. I'll also add a link to some work I and some colleagues are doing on designing social algorithms that promote collective intelligence. https
100.
by jwarden 3y ago
Author here. Thank you. And good idea, it would definitely improve the article to pull out actually examples. I could show examples of righ-wing+helpful, left-wing+helpful, right-wing+not-helpful, neutral+helpful, etc. I have looked at many
101.
by jwarden 3y ago
Yes just one. Adding more latent factors seems to make little difference. A single factor seems to explain most of the polarization among users. I have run it with multiple dimensions and plotted the results. They are fascinating, even if t
102.
Understanding bridge-based ranking(jonathanwarden.com)
66 points by jwarden 3y ago|21 comments
103.
by jwarden 3y ago
I sat in a cafe overlooking a busy street, and pressed f when a man walked by and d when a woman walked by, and got 51%.
104.
by jwarden 3y ago
The claim that the "defund the police" movement had any goal other than defunding the police is an example of "sanewashing": https://www.urbandictionary.com/define.php?term=sanewashing
105.
by jwarden 3y ago
It’s like the old “5 Whys” technique for getting at the root of a problem. But as you suggest, there may not be a root: we can always push “real” causality one level down. Whys all the way down. But that doesn’t mean we can’t answer the que
106.
by jwarden 3y ago
This doesn't seem all that lopsided. A quick rule of thumb I have is that an 80/20 rule translates to the 50/1 rule after repeated application. For example, if 20% of a population has 80% of the wealth, and 20% of 20% of the
107.
by jwarden 3y ago
> Why is surprisal defined as `log (1/p(X=x)`? Good question. It's because -log P(X) is the number of bits of information I would need to eliminate any surprise about the value of X. For example, let's say X is the value o
108.
by jwarden 3y ago
Thank you, fixed.
109.
by jwarden 3y ago
Here's how I describe KL Divergence, building up from simple to complex concepts. surprisal: how surprised I am when I learn the value of X Suprisal(x) = -log p(X=x) entropy: how surprised I expect to be H(p) = 𝔼_X -l
110.
by jwarden 3y ago
The article argues that supreme court justices, like all human beings, have cognitive bias. It does not follow that the supreme court is “broken”. If you accept this argument you must accept all institutions as irretrievably broken. Further
111.
by jwarden 3y ago
Extremely abstruse.
112.
by jwarden 3y ago
I think it's possible. We're working on the problem of protocol-level solutions here: https://social-protocols.org/
113.
by jwarden 3y ago
I think so. I work with a group that was researching ways to improve the hacker news ranking algorithm. We concluded that there were "false positives" and "false negatives" -- stories often did or did not make the home p
114.
Functional Equality: Why 2+2 does not equal 4.0(jonathanwarden.com)
1 points by jwarden 3y ago|0 comments
115.
by jwarden 3y ago
Interesting: once you nail down any two axises, the third follows.
116.
The Law of Attention(deliberati.io)
1 points by jwarden 4y ago|0 comments
117.
by jwarden 4y ago
Nice. It took about a minute to clone it, run it, enter my API key, and get started. The speech-to-text worked flawlessly. Most people can talk faster than they can type, but they can read faster than other people can talk. So an interface
118.
by jwarden 4y ago
Negative equity capital simply means that on the Fed's balance sheet, equity (assets - liabilities) is negative. The Fed's operates on the same fundamental accounting principles as any other organization, so it has assets, liabili
119.
by jwarden 4y ago
Yeah we have thought of a model where Karma is a measure of how much "marginal value" you have created for other users by your submissions. If we take upvotes as a proxy for value, we can calculate how many additional site-wide
120.
by jwarden 4y ago
Yes we'd like to do that, but the main problem is that we don't have access to some data that goes into HN rankings: flags, penalties, and other moderator actions. These make a huge difference to the quality of front-page content.

More ›