7 ms·

I think it's complete bullshit and I fell for it. sigh As a matter of fact, both Hackernews and reddit accounts have wildly different results. Also, the respon

I think it's complete bullshit and I fell for it. sigh As a matter of fact, both Hackernews and reddit accounts have wildly different results. Also, the response is way to quick to fetch all comments from either platform (and you're not telling me you scraped reddit completely, don't you?). Seems to be hugged to death now. So I cannot try with a nonexistent account id.

ben174 4y ago
This is nothing more impressive than the horoscope in the newspaper. It's a cold read and people will always fall for it because they believe what they want to hear.
- Mordisquitos 4y ago
  I find it even less impressive than most horoscopes. At least horoscopes skillfully make use of the Barnum effect[0] to trick people into believing the reading. The results of this on my own users, on the other hand, are much too specific to pretend to hide how wildly inaccurate they are. [0] https://en.wikipedia.org/wiki/Barnum_effect https://en.wikipedia.org/wiki/Barnum_effect
super256 4y ago
> Also, the response is way to quick to fetch all comments from either platform I didn't try the Reddit thingy, but for HN there is the algolia index, which is close to instant. https://hn.algolia.com/?dateRange=all&page=0&prefix=true&query=super256&sort=byPopularity&type=comment https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
- muttled 4y ago
  This is correct. I'm able to grab most users' comment histories in under a second and it tells the data sources at the bottom.
5350-uiop-1130 4y ago
> both Hackernews and reddit accounts have wildly different results why is that unusual, HN is more focused so the comments trend towards work, whereas Reddit is a lot more general.
- rob74 4y ago
  If HN was so focused, that one guy wouldn't have founded the even more "computing-focused" lobste.rs. Mind you, I like it this way, it gives me "food for thought" in both computing and non-computing related fields...
tomthe 4y ago
For both Reddit and HN there are complete dumps that you can download easily within much less than a day. I have worked with both, the reddit dump from pushshift ist quite big (https://files.pushshift.io/reddit/ https://files.pushshift.io/reddit/ several TB?). Scraping HN completely from the API is much, much smaller, around a few GB if I remember correctly. But of course this is still bullshit and doesn't really use that data.
- fragmede 4y ago
  why do you say that?
nzealand 4y ago
I switch reddit accounts every year. It gave mostly consistent results for my last five accounts.
sangnoir 4y ago
I think there's a lot of value in harvesting usernames of active HNers. Maybe I'm being paranoid/cynical, but I think y'all just got Cambridge-Analytica'd
- muttled 4y ago
  You guys are vastly overestimating how much thought I put into this. I cranked it out in a week on PTO.
  - sangnoir 4y ago
    From the outside, it's indistinguishable from Cambridge Analytica. For extra irony: Cambridge Analytica's quiz masquaraded as a personality test[1]. 1. https://www.businessinsider.com/facebook-personality-test-cambridge-analytica-data-trump-election-2018-3 https://www.businessinsider.com/facebook-personality-test-ca...

ben174 4y ago

This is nothing more impressive than the horoscope in the newspaper. It's a cold read and people will always fall for it because they believe what they want to hear.

Mordisquitos 4y ago
I find it even less impressive than most horoscopes. At least horoscopes skillfully make use of the Barnum effect[0] to trick people into believing the reading. The results of this on my own users, on the other hand, are much too specific to pretend to hide how wildly inaccurate they are. [0] https://en.wikipedia.org/wiki/Barnum_effect https://en.wikipedia.org/wiki/Barnum_effect

super256 4y ago

> Also, the response is way to quick to fetch all comments from either platform I didn't try the Reddit thingy, but for HN there is the algolia index, which is close to instant. https://hn.algolia.com/?dateRange=all&page=0&prefix=true&query=super256&sort=byPopularity&type=comment https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

muttled 4y ago
This is correct. I'm able to grab most users' comment histories in under a second and it tells the data sources at the bottom.

5350-uiop-1130 4y ago

> both Hackernews and reddit accounts have wildly different results why is that unusual, HN is more focused so the comments trend towards work, whereas Reddit is a lot more general.

rob74 4y ago
If HN was so focused, that one guy wouldn't have founded the even more "computing-focused" lobste.rs. Mind you, I like it this way, it gives me "food for thought" in both computing and non-computing related fields...

tomthe 4y ago

For both Reddit and HN there are complete dumps that you can download easily within much less than a day. I have worked with both, the reddit dump from pushshift ist quite big (https://files.pushshift.io/reddit/ https://files.pushshift.io/reddit/ several TB?). Scraping HN completely from the API is much, much smaller, around a few GB if I remember correctly. But of course this is still bullshit and doesn't really use that data.

fragmede 4y ago
why do you say that?

nzealand 4y ago

I switch reddit accounts every year. It gave mostly consistent results for my last five accounts.

sangnoir 4y ago

I think there's a lot of value in harvesting usernames of active HNers. Maybe I'm being paranoid/cynical, but I think y'all just got Cambridge-Analytica'd

muttled 4y ago
You guys are vastly overestimating how much thought I put into this. I cranked it out in a week on PTO.
- sangnoir 4y ago
  From the outside, it's indistinguishable from Cambridge Analytica. For extra irony: Cambridge Analytica's quiz masquaraded as a personality test[1]. 1. https://www.businessinsider.com/facebook-personality-test-cambridge-analytica-data-trump-election-2018-3 https://www.businessinsider.com/facebook-personality-test-ca...