manx

31 ms·

331.
by manx 5y ago
Looks good to me! How does your ranking algorithm work? I recently wrote an article about how it works at Hacker News, if you want some food for thought: https://felx.me/2021/08/29/improving-the-hacker-news-ra
332.
by manx 5y ago
There's some luck involved. I explained how it works in a recent article: https://felx.me/2021/08/29/improving-the-hacker-news-ranking...
333.
by manx 5y ago
You are right. Our assumption for now was that clickthroughs are distributed like upvotes. But we don't have any data backing this up. That's why we need to do real experiments. And this touches to your second point that even on t
334.
by manx 5y ago
Now I understand. You're right, that's a real disadvantage. We'll take that into account in our coming research. Thank you!
335.
by manx 5y ago
Doing it client-side could actually work.
336.
by manx 5y ago
Super interesting! I'd love to see what you come up with.
337.
by manx 5y ago
You raised some important points. The threshold of 2 upvotes is a necessary condition to be considered for the front-page (all pages, not just rank 1-30), not a sufficient one. I have this information from dang himself. Garbage on the new-p
338.
by manx 5y ago
AFAIK things like these have been tried, but I can't find dangs comment about it. Maybe he can report about these experiments himself.
339.
by manx 5y ago
Absolutely true. We thought about it, but didn't have any idea yet. Maybe there's also a balancing feedback loop approach to it. Something like 'longer click to upvote delay' should lead to 'higher rank'.
340.
by manx 5y ago
That would be great. Maybe there's a hacker news clone we could adapt. In any case, our resources are limited, so any help is appreciated :) Another option would be an experiment on HN itself for a week or so.
341.
by manx 5y ago
That's a good question. Unfortunately, we don't have any data about the heterogeneity of the user pool. So we have to guess. In the simplified simulation, the agents have random expectation levels and compare that to the submissio
342.
by manx 5y ago
I'm thinking again about your question. I think you're talking about the time span a submission is visible on the frontpage, right? This can be controlled by the age exponent in the formula and (in our proposal) by the upvote-thre
343.
by manx 5y ago
Which data or table are you referring to? "Later" in the list of scores when submitted multiple times (first table)? Note that this is a 30-day window. Growth of the whole community can be seen in the kaggle notebook: https:&#x2F
344.
by manx 5y ago
I'm sorry that wasn't clear enough. Our current working definition can be described as: We define a view as a click by a registered user on a submission, so the user sees the submitted content. For Ask HN, the content would be the
345.
by manx 5y ago
I did. dang said I should post it and let the community discuss about it.
346.
by manx 5y ago
Our simple simulation had a 1-second time resolution, so it evaluated the formula for all link positions over the submission's lifetime. But there was no pagination. We're working on a more accurate simulation that will have many
347.
Improving the Hacker News Ranking Algorithm(felx.me)
122 points by manx 5y ago|54 comments
348.
What's the Maturity Level of ScalaJS?(reddit.com)
1 points by manx 5y ago|0 comments
349.
by manx 5y ago
What is it for? Is it just a chat?
350.
Ask HN: What are current good Linux developer laptops?
16 points by manx 5y ago|7 comments
351.
by manx 5y ago
Could you elaborate?
352.
Ask HN: What kind of like-minded people are you looking for?
11 points by manx 5y ago|22 comments
353.
by manx 5y ago
No real users. It's a simplified simulation of a frontpage and how random users interact with it. Have a look at the code: https://github.com/fdietze/downvote-scoring/blob/master/scor... I started t
354.
by manx 5y ago
With a different ranking formula it's actually quite promising. I created a simple simulation and some plots: https://github.com/fdietze/downvote-scoring
355.
by manx 5y ago
I'm researching different voting algorithms for HN. If you're interested, please contact me (email in profile).
356.
by manx 5y ago
A bit more information for your mental model: Afaik, reddit has a pretty stable upvote vs downvote ratio, like 7:1. (Sorry, no source, but I remember the reddit founder saying it in an interview). And in your calculations you assumed that e
357.
by manx 5y ago
I agree that we need negative feedback in our voting systems. But it doesn't have to be downvotes. You can use the number of views as a negative force. I created some simulations and plots: https://github.com/fdietze&#x
358.
by manx 5y ago
I recommend adding a little trick to ignore existing ~/.sqliterc files, which makes the script more portable: cat <<EOF | sqlite3 -init <(echo "") "$DBFILE" ...your sql... EOF For one-line
359.
by manx 5y ago
I'm not talking about the intelligence of some individuals, but of groups acting as an entity. Let's say aliens look at earth and ask why earthlings don't do anything against their extinction. If the answer they get is "
360.
by manx 5y ago
We're pairing on our daily work. So everybody learns about the technologies, tooling, specific parts of a codebase and mindset the other person is working with.

More ›