New weekly digest feature - summary of 10 featured articles and 252 total articles from newsletters written by people in the SheWritesAI directory in the past 7 days
Thank you for including me in your list. I’m just curious about how the score is being calculated, could you please share more details on the methodology? Also, do you consider it a metric that can be improved over time? 🙈
Hi Farida, that's a great question for StackDigest creator Karen Spinner 😊 My understanding is it's currently a balance of length and engagement. I believe engagement includes likes and comments, with comments weighing twice as much as a like. I think restacks are not currently being counted. Karen can explain more.
The scoring model treats comments and likes equally and uses a progressive scale with soft caps that weights the first 10 engagements highest with gradually diminishing returns through 100. The idea behind this is to slightly disadvantage articles with very high engagement and help surface “hidden gems.” I need to update the FAQ with a more detailed description of this algo. I’m also thinking about adding an option that will allow users to set relative weights for likes and comments.
Restacks aren’t included because they can reflect how intensely authors promote their work rather than community conversations. The scoring model will be revised soon (hopefully this week) to use average hourly engagement (engagement total/hours since publication) for more equitable scoring.
I’m also looking into how this figure might be further adjusted to factor in number of subscribers, since that has a big impact on engagement as well!
Thanks for the mention, Karen. So many interesting articles here I can’t wait to dig into!
Thank you for including me on your list!
Thank you for including me in your list. I’m just curious about how the score is being calculated, could you please share more details on the methodology? Also, do you consider it a metric that can be improved over time? 🙈
Hi Farida, that's a great question for StackDigest creator Karen Spinner 😊 My understanding is it's currently a balance of length and engagement. I believe engagement includes likes and comments, with comments weighing twice as much as a like. I think restacks are not currently being counted. Karen can explain more.
Hi Farida and Karen,
The scoring model treats comments and likes equally and uses a progressive scale with soft caps that weights the first 10 engagements highest with gradually diminishing returns through 100. The idea behind this is to slightly disadvantage articles with very high engagement and help surface “hidden gems.” I need to update the FAQ with a more detailed description of this algo. I’m also thinking about adding an option that will allow users to set relative weights for likes and comments.
Restacks aren’t included because they can reflect how intensely authors promote their work rather than community conversations. The scoring model will be revised soon (hopefully this week) to use average hourly engagement (engagement total/hours since publication) for more equitable scoring.
I’m also looking into how this figure might be further adjusted to factor in number of subscribers, since that has a big impact on engagement as well!