Fix SQL for @lastwhale twitter bot

[zimmeee]

I think it's not right

[wasauce]

What should it be doing? --  Bill Ferrell 650.646.2065 From: Noah Zimmerman notifications@github.com Reply: zimmeee/fortunatepun reply@reply.github.com Date: February 12, 2014 at 3:44:16 PM To: zimmeee/fortunatepun fortunatepun@noreply.github.com Subject:  [fortunatepun] Fix SQL for @lastwhale twitter bot (#16)
I think it's not right

— Reply to this email directly or view it on GitHub.

[zimmeee]

It's tweeting, but I don't think it's counting the link votes properly, which would be my fault. I think it's not counting correctly because my page, the top link has 5 votes, so I think that is likely to be the top link, but it is not what the bot is twitting https://twitter.com/lastwhale/status/433670781290029056

I'm pretty sure I know what is wrong.

On Feb 12, 2014, at 3:46 PM, wasauce notifications@github.com wrote:

What should it be doing?

Bill Ferrell 650.646.2065 From: Noah Zimmerman notifications@github.com Reply: zimmeee/fortunatepun reply@reply.github.com Date: February 12, 2014 at 3:44:16 PM To: zimmeee/fortunatepun fortunatepun@noreply.github.com Subject: [fortunatepun] Fix SQL for @lastwhale twitter bot (#16) I think it's not right

— Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub.

[zimmeee]

Ok, I was in here ready to fix this and then I realized I'm not sure what the intended behavior should be. That code that I just wrote gets the most popular links in the last hour. Then the bot would tweet the top link of the hour. But that's actually pretty boring unless we had a shitload of users, since most URLs only get 1 or 2 votes in that time. Alternatively, it could function just like our newsfeed page and aggregate the URLs over a 24 hour period, except for the bot it would aggregate over ALL users. So then the bot is tweeting a rolling average of the top stories in the past 24 hours. I think this is better in the short-term, though it will definitely lead to some repeat link posting unless we put in some additional logic.

Thoughts?

[wasauce]

Twitter won’t repeat the same link — which is why our twitter feed appears to update sporadicly — so I don’t think we need any more logic… but I do agree rolling 24 hours sounds great! --  Bill Ferrell 650.646.2065 From: Noah Zimmerman notifications@github.com Reply: zimmeee/fortunatepun reply@reply.github.com Date: February 12, 2014 at 7:17:13 PM To: zimmeee/fortunatepun fortunatepun@noreply.github.com Cc: wasauce wferrell@gmail.com Subject:  Re: [fortunatepun] Fix SQL for @lastwhale twitter bot (#16)
Ok, I was in here ready to fix this and then I realized I'm not sure what the intended behavior should be. That code that I just wrote gets the most popular links in the last hour. Then the bot would tweet the top link of the hour. But that's actually pretty boring unless we had a shitload of users, since most URLs only get 1 or 2 votes in that time. Alternatively, it could function just like our newsfeed page and aggregate the URLs over a 24 hour period, except for the bot it would aggregate over ALL users. So then the bot is tweeting a rolling average of the top stories in the past 24 hours. I think this is better in the short-term, though it will definitely lead to some repeat link posting unless we put in some additional logic.

Thoughts?

— Reply to this email directly or view it on GitHub.

[zimmeee]

Pushed fix 1f34ff3b72bc6ba152c33cad0b92d61180418c35 I also added the twitter handles of the people who tweeted the top URL in the result. I thought maybe the bot could @ those twats in it's broadcast to really get our viral numbers cranking. Such virus. Many spam.

[rbruggner]

I currently support top link aggregated across all users for 24hr time period. This can be biased by one individual (if they follow many people) or a set of people (if Noah and I follow many of the same users).

Wait. Now I don't support that.

This could be fixed by normalizing by the number of users an individual follows.

Alternatives: Most frequently followed link. I think I like that. But that's not relevant until we get more users.

Can we make subject area bots that follow a curated set of influential people? Then the subject bots can tweet their top links. We could hand these out to people who know about a subject area.

All of these suggestions take time. I need a burrito.

On Feb 12, 2014, at 7:18 PM, wasauce notifications@github.com wrote:

Twitter won’t repeat the same link — which is why our twitter feed appears to update sporadicly — so I don’t think we need any more logic… but I do agree rolling 24 hours sounds great!

Bill Ferrell 650.646.2065 From: Noah Zimmerman notifications@github.com Reply: zimmeee/fortunatepun reply@reply.github.com Date: February 12, 2014 at 7:17:13 PM To: zimmeee/fortunatepun fortunatepun@noreply.github.com Cc: wasauce wferrell@gmail.com Subject: Re: [fortunatepun] Fix SQL for @lastwhale twitter bot (#16) Ok, I was in here ready to fix this and then I realized I'm not sure what the intended behavior should be. That code that I just wrote gets the most popular links in the last hour. Then the bot would tweet the top link of the hour. But that's actually pretty boring unless we had a shitload of users, since most URLs only get 1 or 2 votes in that time. Alternatively, it could function just like our newsfeed page and aggregate the URLs over a 24 hour period, except for the bot it would aggregate over ALL users. So then the bot is tweeting a rolling average of the top stories in the past 24 hours. I think this is better in the short-term, though it will definitely lead to some repeat link posting unless we put in some additional logic.

Thoughts?

— Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub.

[zimmeee]

Interesting you bring that up. The original lastwhalestanding, lonelygirl32.com, was just that: a domain specific version with a curated list of scientist. so yeah, i like the domain-bot idea. i think an easy path to implementation is to piggyback on Twitter's list - people are already investing time/effort in building them. I could imagine feeding a handle + list and generating a customized feed.

p.s. early mockup of lonelygirl, aka lastwhalestanding 1.0

On Feb 12, 2014, at 7:33 PM, Robert Bruggner notifications@github.com wrote:

I currently support top link aggregated across all users for 24hr time period. This can be biased by one individual (if they follow many people) or a set of people (if Noah and I follow many of the same users).

Wait. Now I don't support that.

This could be fixed by normalizing by the number of users an individual follows.

Alternatives: Most frequently followed link. I think I like that. But that's not relevant until we get more users.

Can we make subject area bots that follow a curated set of influential people? Then the subject bots can tweet their top links. We could hand these out to people who know about a subject area.

All of these suggestions take time. I need a burrito.

On Feb 12, 2014, at 7:18 PM, wasauce notifications@github.com wrote:

Twitter won’t repeat the same link — which is why our twitter feed appears to update sporadicly — so I don’t think we need any more logic… but I do agree rolling 24 hours sounds great!

Bill Ferrell 650.646.2065 From: Noah Zimmerman notifications@github.com Reply: zimmeee/fortunatepun reply@reply.github.com Date: February 12, 2014 at 7:17:13 PM To: zimmeee/fortunatepun fortunatepun@noreply.github.com Cc: wasauce wferrell@gmail.com Subject: Re: [fortunatepun] Fix SQL for @lastwhale twitter bot (#16) Ok, I was in here ready to fix this and then I realized I'm not sure what the intended behavior should be. That code that I just wrote gets the most popular links in the last hour. Then the bot would tweet the top link of the hour. But that's actually pretty boring unless we had a shitload of users, since most URLs only get 1 or 2 votes in that time. Alternatively, it could function just like our newsfeed page and aggregate the URLs over a 24 hour period, except for the bot it would aggregate over ALL users. So then the bot is tweeting a rolling average of the top stories in the past 24 hours. I think this is better in the short-term, though it will definitely lead to some repeat link posting unless we put in some additional logic.

Thoughts?

— Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub.

[rbruggner]

How early?

On Feb 12, 2014, at 9:58 PM, Noah Zimmerman notifications@github.com wrote:

Interesting you bring that up. The original lastwhalestanding, lonelygirl32.com, was just that: a domain specific version with a curated list of scientist. so yeah, i like the domain-bot idea. i think an easy path to implementation is to piggyback on Twitter's list - people are already investing time/effort in building them. I could imagine feeding a handle + list and generating a customized feed.

p.s. early mockup of lonelygirl, aka lastwhalestanding 1.0

On Feb 12, 2014, at 7:33 PM, Robert Bruggner notifications@github.com wrote:

I currently support top link aggregated across all users for 24hr time period. This can be biased by one individual (if they follow many people) or a set of people (if Noah and I follow many of the same users).

Wait. Now I don't support that.

This could be fixed by normalizing by the number of users an individual follows.

Alternatives: Most frequently followed link. I think I like that. But that's not relevant until we get more users.

Can we make subject area bots that follow a curated set of influential people? Then the subject bots can tweet their top links. We could hand these out to people who know about a subject area.

All of these suggestions take time. I need a burrito.

On Feb 12, 2014, at 7:18 PM, wasauce notifications@github.com wrote:

Twitter won’t repeat the same link — which is why our twitter feed appears to update sporadicly — so I don’t think we need any more logic… but I do agree rolling 24 hours sounds great!

Bill Ferrell 650.646.2065 From: Noah Zimmerman notifications@github.com Reply: zimmeee/fortunatepun reply@reply.github.com Date: February 12, 2014 at 7:17:13 PM To: zimmeee/fortunatepun fortunatepun@noreply.github.com Cc: wasauce wferrell@gmail.com Subject: Re: [fortunatepun] Fix SQL for @lastwhale twitter bot (#16) Ok, I was in here ready to fix this and then I realized I'm not sure what the intended behavior should be. That code that I just wrote gets the most popular links in the last hour. Then the bot would tweet the top link of the hour. But that's actually pretty boring unless we had a shitload of users, since most URLs only get 1 or 2 votes in that time. Alternatively, it could function just like our newsfeed page and aggregate the URLs over a 24 hour period, except for the bot it would aggregate over ALL users. So then the bot is tweeting a rolling average of the top stories in the past 24 hours. I think this is better in the short-term, though it will definitely lead to some repeat link posting unless we put in some additional logic.

Thoughts?

— Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub. — Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[zimmeee]

8/19/2010 !!!

[zimmeee]

I don't see the image in this github thread - did you see it? Jonathan Eisen? hah

[zimmeee]

[rbruggner]

Ahhh.. that’s better. I was just sad that lonelygirl32.com wasn’t live.

It looks a lot like the whale :)

On Feb 12, 2014, at 10:07 PM, Noah Zimmerman notifications@github.com wrote:

— Reply to this email directly or view it on GitHub.

[zimmeee]

2010: 50m tweets/day 2014: 500m tweets/day

That's less gain than I would have guessed