You are viewing a single comment's thread from:

RE: Introducing "Steem Follower Checker": A new Open Source browser plugin

in Steem Dev • 7 months ago

Very interesting.
The new constraints and limits seem to be very sensible. I did see your changes (and the new heatmap) after my comment last week, but unfortunately didn't have time to look at it in more detail.
I see you've made good progress though. I'm curious and will definitely try it out...

7 months ago in Steem Dev by
$9.44
  • Past Payouts $9.44
  • - Author $4.72
  • - Curators $4.72
8 votes
Reply 11
Sort:  
  • Trending
    • [-]
     7 months ago 

    Now that it's easier to check peoples' scores in the browser, I'm seeing that it's probably far too easy to get a top score. I'm thinking I'll start planning for a June 30 methodology change, then tentatively once a quarter after that. I think the first order of business for June is to make it harder to max the score out.

    $6.29
    • Past Payouts $6.29
    • - Author $3.15
    • - Curators $3.15
    3 votes
    Reply
    [-]
     7 months ago 

    Yes, a few adaptations will certainly still be necessary. Fine adjustment is usually not that easy and takes time. I haven't looked closely yet, but I couldn't say right off the bat how you could easily get to the top.

    Integration in the Browser Extension would be helpful to be able to track the results better. However, the additional requests prevent me from doing this. Basically, it would be better if the follower median was already calculated at hivemind level. That shouldn't be so much effort there. Maybe I can integrate this as a test (if the time allows it).

    $0.06
    • Past Payouts $0.06
    • - Author $0.03
    • - Curators $0.03
    2 votes
    Reply
    [-]
     7 months ago 

    Integration in the Browser Extension would be helpful to be able to track the results better. However, the additional requests prevent me from doing this.

    This was exactly my thinking. I wanted to add word counts, too, but after thinking it through I thought it would be too much network traffic. Now I'm thinking of maybe showing both values one post/author at a time via mouseover and/or after clicking into a post.

    Basically, it would be better if the follower median was already calculated at hivemind level. That shouldn't be so much effort there. Maybe I can integrate this as a test (if the time allows it).

    That would definitely be a useful addition. I have no idea about the work effort, though. If it's not already there, the follower count would be useful, too. A lot of the API servers aren't running the follower plugin.

    $0.22
    • Past Payouts $0.22
    • - Author $0.11
    • - Curators $0.11
    2 votes
    Reply
    [-]
     7 months ago 

    Now I'm thinking of maybe showing both values one post/author at a time via mouseover and/or after clicking into a post.

    Mouseover would be conceivable, but I'm not sure if the data could be loaded (and processed) that quickly.
    I can well imagine the information for the post display. The word count and the number of followers could be extracted from the data already available in the Redux Store. We wouldn't need an additional query for this. However, I don't know whether the extension can access the Redux Store. I could imagine that this is not so easily possible.

    A lot of the API servers aren't running the follower plugin

    The plugin is not necessary for this. This data is also stored in the hivemind database. A (not so simple) SQL query is sufficient for this (The copilot helped with the median function):

    SELECT PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY accounts.reputation) AS median_reputation
FROM hive_follows AS follows
JOIN hive_accounts AS accounts ON follows.follower = accounts.id
WHERE follows.following = :aid AND follows.state IN (1,3);

    The calculated values were exactly the same as those I had determined with my last test code.

    $0.18
    • Past Payouts $0.18
    • - Author $0.09
    • - Curators $0.09
    3 votes
    Reply
    [-]
     7 months ago 

    The word count and the number of followers could be extracted from the data already available in the Redux Store. We wouldn't need an additional query for this. However, I don't know whether the extension can access the Redux Store. I could imagine that this is not so easily possible.

    There's a Google Chrome Redux Plugin which can be used for debugging - so presumably, it's possible for a different Plugin to access the Redux state within a website.

    $0.10
    • Past Payouts $0.10
    • - Author $0.05
    • - Curators $0.05
    1 vote
    Reply
    [-]
     7 months ago 

    Do you mean the Redux DevTools? As far as I know, the DevTools should be deactivated in the production environment.

    Or is there another plugin?

    $0.06
    • Past Payouts $0.06
    • - Author $0.03
    • - Curators $0.03
    2 votes
    Reply
    [-]
     7 months ago 

    Ah, yes. I do mean the Redux dev tools. I’ve only used them in test so it sounds like you can dismiss that idea.

    $0.10
    • Past Payouts $0.10
    • - Author $0.05
    • - Curators $0.05
    1 vote
    Reply
    [-]
     7 months ago 

    The plugin is not necessary for this. This data is also stored in the hivemind database. A (not so simple) SQL query is sufficient for this (The copilot helped with the median function):

    So, the way that works is that the SQL query gets added to the bridge API in hivemind, and then the browser extension would make a bridge API call to a steem node? Would a soft fork be needed? This is still sort of a black box to me.

    (and from above)

    a few adaptations will certainly still be necessary

    I think I have a potentially interesting and impactful idea for the June 30 update. I'll try to post about it this weekend so I can get community feedback before implementing.

    $0.10
    • Past Payouts $0.10
    • - Author $0.05
    • - Curators $0.05
    1 vote
    Reply
    [-]
     7 months ago 

    added to the bridge API in hivemind, and then the browser extension would make a bridge API call to a steem node?

    Right. Like the new get_bookmarked_posts here

    Would a soft fork be needed?

    No, only an update of hivemind nodes. We don't even have to wait for Steemit to integrate this into their nodes, as we can also use the extension to query our own nodes (e.g. the one on which my database version 21 with the bookmarks is already running).

    $0.06
    • Past Payouts $0.06
    • - Author $0.03
    • - Curators $0.03
    3 votes
    Reply
    [-]
     7 months ago 

    It hadn't crossed my mind before, but this morning I wanted to see if I could merge the content extension and the popup extension into one, and if so, how hard it would be. It turned out to be fairly easy, and took less than an hour. So now I need to decide whether there should be one extension or two... I guess ideally, there would be a single extension with:

    • a configuration capability for the user to activate either or both modules; and
    • a shared library so that both modules could share functions like the follower network scoring.

    That's all over my head right now, though.

    $0.10
    • Past Payouts $0.10
    • - Author $0.05
    • - Curators $0.05
    1 vote
    Reply
    [-]
     7 months ago 

    I've already read your post on this, but won't be able to reply to you today.

    $0.06
    • Past Payouts $0.06
    • - Author $0.03
    • - Curators $0.03
    2 votes
    Reply