Timeline for If more users could vote, would they engage more? Testing 1 reputation voting on some sites
Current License: CC BY-SA 4.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 2, 2023 at 16:01 | comment | added | Catija StaffMod | But in many of these cases, we'll want to investigate where these votes are coming from to better understand the motivation for these votes. Other possible outcomes are ones we see throughout this discussion - unexpected numbers of socks being created to target votes at specific accounts (up or down). It can seem somewhat hand-wavy because we understand that specific sites will have different situations that are concerning to them and we want to lean on the expertise of the site mods to understand what risks they see. | |
| Oct 2, 2023 at 15:53 | comment | added | Catija StaffMod | There could be positive or negative outcomes we'll only see in six months or a year because this change leads to a gradual shift. So, again, rolling back and considering the test "done" after only a month would negatively impact the long-term understanding of the change. When it comes to negative outcomes, we're thinking of huge increases in moderator flags or need to handle content, or the community indicating that they can't delete content that they should be able to due to an influx in upvotes on good content... or good content being downvoted when it shouldn't. | |
| Oct 2, 2023 at 15:49 | comment | added | Catija StaffMod | @brhans I don't think that's quite correct. As my comments say, the goal is to avoid granting, removing and re-granting privileges in a short period of time. There is the potential that we'll look at the data after the test period and determine that it's causing more harm than good and roll it back at that point. I don't mention that, so it's understandable why you'd interpret it that way. The other thing worth keeping in mind is that this sort of change may not be immediately understandable. Removing the test after only a month could cause us to overlook much. | |
| Oct 2, 2023 at 14:40 | comment | added | user400654 | @brhans presumably that'd mean an unmitigable amount of fraud being caused by it, since that is the primary concern for the outcome of this change. Given it's going to roll out to select sites, not network-wide, during the experiment we should get a fairly effective view from that as far as what additional tools are necessary to mitigate any increase in fraud before it rolls out to larger sites. Put another way, a negative outcome isn't necessarily the end of the experiment, rather, it's an opportunity to mitigate those problems and continue. | |
| Oct 2, 2023 at 13:11 | comment | added | brhans | If only "unexpected negative outcomes" might lead to a roll-back of this 'experiment', could you clarify what some expected negative outcomes might be, and why these negative outcomes would be acceptable and not real to a roll-back of the 'experiment'? | |
| Oct 2, 2023 at 13:10 | comment | added | brhans | @Catija the way you describe the roll-back plan here reads to me as if SE has already decided to implement this plan come hell or high-water, but is 'whitewashing' it by announcing it as an experiment to make it look as though you're acting in good faith. The lack of a defined period of time over which this 'experiment' will run further reinforces this impression for me. In my mind, an experiment should run for some defined period of time while you gather data. After the experiment ends you make a decision - using the data you gathered - whether to implement the thing you experimented with. | |
| Sep 20, 2023 at 3:27 | comment | added | Catija StaffMod | The main line of reasoning for not rolling back at the end of the test period is because we're going to be reviewing data throughout the process. If there's no overt negative impact we recognize during the test, it's unclear that rolling back while reviewing additional data is necessary. This also risks the impacted low-rep users being confused by the loss of the voting privilege they just gained... which might be reinstated again a short time later. This is a change from other tests we've run in the past but I think it makes sense in this situation. | |
| Sep 20, 2023 at 3:16 | comment | added | Catija StaffMod | The main change of a rollback would be resetting the privilege levels for up- and downvoting to their current levels (15 and 125) to prevent additional voting. In my understanding, we likely wouldn't remove all low-reputation votes cast during the test by default. Our general expectation is that most votes cast would be within site expectations so we'd focus on invalidating votes known to be problematic. That doesn't mean we wouldn't ever decide to remove all votes but doing so would likely be quite complex as votes give rep and rep gives voting... someone can earn the privilege mid-test. | |
| Sep 20, 2023 at 2:59 | comment | added | Catija StaffMod | Our plan is to only roll back if we meet the guidance for doing so or if unexpected negative outcomes cause us to determine that rolling back is the best course of action. This can happen at any point during the test, not only after the test period has elapsed. If we can make adjustments or improvements to tools instead of rolling back, that's an alternative we will consider. We'd prefer finding ways to address issues rather than having to stop the test entirely but, in particular, if addressing issues would take too long, we will revert the change. | |
| Sep 20, 2023 at 0:42 | history | answered | starballMod | CC BY-SA 4.0 |