Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
c_prompt
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
11 ms
·
61.
▲
by
c_prompt
11y ago
Do you have any plans to create something similar for renewals?
62.
▲
by
c_prompt
11y ago
tl/dr: Feeding America [1] is a network of food banks that feeds more than 46 million people. In 2005, four professors at the University of Chicago helped replace their centralized distribution system with an auction-based one, allocat
63.
▲
Success: what happens when America's food banks embrace free-market economics [pdf]
(conference.nber.org)
2 points
by
c_prompt
11y ago
|
1 comments
64.
▲
Antidepressant was misrepresented as safe for adolescents
(adelaide.edu.au)
5 points
by
c_prompt
11y ago
|
0 comments
65.
▲
by
c_prompt
11y ago
This isn't a direct answer to your questions but I hate this disease with a passion and any discussions about mainstream research avenues that ignore other less well-accepted forms of treatment get my emotions in a whirl. (My sister di
66.
▲
by
c_prompt
11y ago
Conceptually, I agree with his view that without a better understanding and appreciation for philosophy, humanity is limited by the progress science makes. But if I'm understanding his reasoning, he wants more philosophy so that scient
67.
▲
by
c_prompt
11y ago
Here's one example why this is a bad decision. From a sailing forum [1]: "I just read an article... that says that there will be no way to turn off updates on Windows 10. That's going to impact cruisers with limited access to
68.
▲
Open letter to Steve Huffman, Reddit's new CEO, with an offer to exchange values
(valme.io)
1 points
by
c_prompt
11y ago
|
0 comments
69.
▲
by
c_prompt
11y ago
She actually tried the junior college route when she lived in a college town to finish her degree. Hated it (not to mention the financial burden). I'm frustrated with myself because I'm out of ideas. I know I can't be her int
70.
▲
Ask HN: Give direction to artist for developing technical skills as backup plan
2 points
by
c_prompt
11y ago
|
2 comments
71.
▲
by
c_prompt
11y ago
> The $100 million acquisition proves that there is no more bar for ‘quality content.’ Quality content is only a fraction of the equation for illustrating #value in a media brand. Projecting the idea of having ‘quality content’ might eve
72.
▲
by
c_prompt
11y ago
A few related articles of interest: Why I Am Not a Member of the American Psychological Association - http://www.chalquist.com/apa.html EMAILS REVEAL CLOSE RELATIONSHIP BETWEEN PSYCHOLOGY GROUP AND CIA - https://
73.
▲
by
c_prompt
11y ago
> Allowing mods to tweak the sorting algorithm is a great idea to do this by the way. Already built. The front page algorithm section at valME.io allows you to select reddit's as the default [1] or choose your own customizations (in
74.
▲
by
c_prompt
11y ago
> I honestly don't understand the necessity or want for something like this. Here's what I wrote elsewhere: "Imagine you're a programmer. For us coding geeks, sites like Stack Overflow are a lifeline to getting work d
75.
▲
by
c_prompt
11y ago
> So you're saying if I create two accounts, one that posts illegal content and one that very quickly flags it, I'll gain a bounty from users each time? No. The bounty comes solely from dividing up the upvotes among the users w
76.
▲
by
c_prompt
11y ago
> There is no downvoting in this system, only upvoting. (How would you take bitcoin from someone?) A downvote is a transaction that costs like anything else, but you increase the cost of the downvote. For example, 2 karma to post (given
77.
▲
by
c_prompt
11y ago
> That is true. I don't agree. You're going to have different communities, just as reddit does now. Intellectuals are (likely) not going to be posting in /r/I-love-taylor-swift; they would post to their own communitie
78.
▲
by
c_prompt
11y ago
You could have flags for CP and then download content with a zero CP flag score. It could work similar to centralized spam providers where you check an IP or username before they post or register. If they have a spam score higher than x, yo
79.
▲
by
c_prompt
11y ago
If you charged very small transaction fees (for posts, comments, upvotes) and charged higher fees for downvotes, you wouldn't completely eliminate the problems but you'd reduce them significantly as it would now cost something rea
80.
▲
by
c_prompt
11y ago
> Who would flag it? Can I trust them? >This is the general problem of reputation, trust, authentication, and naming all in one. I think this project should aim to solve these problems sufficiently well to make things work effectively
81.
▲
by
c_prompt
11y ago
I don't want to invite illegal content, but I am strongly opposed to censorship - in any form. So my question around which policies are best is in the context of no censorship unless something is (likely) illegal.
82.
▲
by
c_prompt
11y ago
Having now reviewed your documents on Github, I don't find anything that specifies policies on how questionable content is to be dealt with (in line with your comment of just defining minimum required standards for users). I'm int
83.
▲
by
c_prompt
11y ago
Only (potentially) illegality. For example, although it's not clear if voat received any formal legal notices or threats before they banned many communities, if they banned them based on their desires about what they want voat to be [1
84.
▲
by
c_prompt
11y ago
Thanks for making the T&Cs you created available. Did your attorneys make them specific to an individual country/locale or did they make them generic enough to afford/enhance protection globally? I'm referring to the latt
85.
▲
by
c_prompt
11y ago
I realize I haven't defined "best" or "more effective" and would also be interested in your views for definitional standards. I have not added my opinions as I prefer not to bias your feedback.
86.
▲
Ask HN: Best policies for community websites for handling questionable content?
5 points
by
c_prompt
11y ago
|
11 comments
87.
▲
by
c_prompt
11y ago
"You said it - they stink on ice." [1] Sorry, couldn't help myself. [1] https://youtu.be/sztf4hcGrB4?t=1m30s
88.
▲
by
c_prompt
11y ago
For my site, I coded a recommendation engine based on cosine similarity. I initially tried Bayes' Theorem and conditional probabilities but abandoned the effort when I didn't see results that made a lot of sense (probably due to a
89.
▲
by
c_prompt
11y ago
If you're looking for something different yet similar, I run valME.io which has a very different business model. And because of the business model, it will have significantly less spam and other undesirable content. But the concept it&
90.
▲
In defense of Google
(valme.io)
3 points
by
c_prompt
11y ago
|
0 comments
More ›