I would expect the karma evaluation to be encapsulated. It would need to be rewritten to take a personal configuration that is editable, and then the calculation would need to be changed to run the dot product of the configuration and the votes. That doesn’t seem that hard.
Probably creating the configurable karma weighting would be the majority of the code changes, while the change in calculation would be a few lines.
To clarify, my claim was that it would be really damn hard to do this while keeping page loading times reasonable. I’m less confident of that than I used to be though.
I also object that there are privacy implications; but full disclosure, I think my true rejection is just that this idea strikes me as… ugly, I guess.
Ugly? Customization strikes me as functional and stylish. The privacy implications would be real, but there should be simple ways to mitigate, but not entirely eliminate the issue. People could opt in to allow themselves to be weighted, and maybe there should be a minimum number of non zero weights required.
Opt-in makes it fairly pointless. But without opt-in, if you want to make it prohibitively difficult to write a script to find a member’s complete voting record with reasonable speed and near-total confidence:
I think you would need to make this only available above a certain karma threshold and with a significant time-delay before changes took effect, (so users can’t see a page with arbitrary weights on-demand). And only permit a few thresholds of weighting and require several users at each level (a single user with a fractional weight has no privacy).
Then someone can just weight sockpuppets who have no votes, or members who weren’t active during the time period you’re interested in, or …
And if there’s a way of getting around that, two users collaborating (or one user who’s posted a rationality quote with a sockpuppet) can still blow this out of the water.
And all this makes the feature even uglier than it was before.
As to ugly: this inferential gap is probably larger than I care to bridge.
A dot product is really damned hard?
I would expect the karma evaluation to be encapsulated. It would need to be rewritten to take a personal configuration that is editable, and then the calculation would need to be changed to run the dot product of the configuration and the votes. That doesn’t seem that hard.
Probably creating the configurable karma weighting would be the majority of the code changes, while the change in calculation would be a few lines.
To clarify, my claim was that it would be really damn hard to do this while keeping page loading times reasonable. I’m less confident of that than I used to be though.
I also object that there are privacy implications; but full disclosure, I think my true rejection is just that this idea strikes me as… ugly, I guess.
Ugly? Customization strikes me as functional and stylish. The privacy implications would be real, but there should be simple ways to mitigate, but not entirely eliminate the issue. People could opt in to allow themselves to be weighted, and maybe there should be a minimum number of non zero weights required.
Okay, you just nerd sniped me.
Opt-in makes it fairly pointless. But without opt-in, if you want to make it prohibitively difficult to write a script to find a member’s complete voting record with reasonable speed and near-total confidence:
I think you would need to make this only available above a certain karma threshold and with a significant time-delay before changes took effect, (so users can’t see a page with arbitrary weights on-demand). And only permit a few thresholds of weighting and require several users at each level (a single user with a fractional weight has no privacy).
Then someone can just weight sockpuppets who have no votes, or members who weren’t active during the time period you’re interested in, or …
And if there’s a way of getting around that, two users collaborating (or one user who’s posted a rationality quote with a sockpuppet) can still blow this out of the water.
And all this makes the feature even uglier than it was before.
As to ugly: this inferential gap is probably larger than I care to bridge.