Install counts starting December 1 (UTC) should be about 50% higher than they were previously.
Greasy Fork has measures in place to ensure install counters are accurate to the number of people who actually installed. This is to prevent bots from easily sending a script to the top of the rankings, intentionally or unintentionally. One of the measures involves generating a secret value that changes every 5 minutes. A derivative of this value is calculated and must be included for an install to count. What I found is that Greasy Fork's backend servers were each generating different values and storing them in their separate caches instead of using the same value.
When you browse Greasy Fork, your request is routed to one of the backend servers. Normally it doesn't matter, and normally you stay on the same server you start on. But with this issue, if you end up getting switched a different server, your install doesn't count towards the counter (though you still get to install the script normally).
I've adjusted how this works so that each backend server will each deterministically generate the same value, so even if you switch servers at some point, it will still increase the install counter when you install.
Install counts starting December 1 (UTC) should be about 50% higher than they were previously.
Greasy Fork has measures in place to ensure install counters are accurate to the number of people who actually installed. This is to prevent bots from easily sending a script to the top of the rankings, intentionally or unintentionally. One of the measures involves generating a secret value that changes every 5 minutes. A derivative of this value is calculated and must be included for an install to count. What I found is that Greasy Fork's backend servers were each generating different values and storing them in their separate caches instead of using the same value.
When you browse Greasy Fork, your request is routed to one of the backend servers. Normally it doesn't matter, and normally you stay on the same server you start on. But with this issue, if you end up getting switched a different server, your install doesn't count towards the counter (though you still get to install the script normally).
I've adjusted how this works so that each backend server will each deterministically generate the same value, so even if you switch servers at some point, it will still increase the install counter when you install.