Varying results (up and down) in repeated shared counts Refresh

I’ve noticed that the results of the shared count are not consistent and noticed some for me strange results.

I find if f.i. strange to have a http count on a new blog post that is fairly new and has never existed in a http version. And I expect the same result when I repeatedly refresh the shared counts in the editor.

Instead each time I use the counts refresh button in the editor it changes the count in an way I can’t explain yet. ofcourse there can be some activity between refreshes, but a social share count is not likely to go down in the short time it takes to refresh again. Especially for a post that has not that much activity in the first place.

I investigated if it is the plugin that is causing this, or the sharedcounts.com api and found that it is probably the sharedcounts.com api. So I wonder how reliable the sharedcounts.com information is.

Testing the api call in postman, it at first seemed alright there (each refresh gave me the same result), what made me think the plugin was corrupting the data in some way, but after restarting postman the numbers where different than the previous time, so now I’m thinking that postman gets the repeated results from some sort of cache instead. So now I restart postman between each get.

I tested with this url:

https://api.sharedcount.com/v1.0/?apikey={key}&url=https://dewitteprins.nl/moet-ik-mijn-kind-laten-vaccineren/

Some of the results (only the Facebook part):

"Facebook": {
        "comment_plugin_count": 0,
        "total_count": 2,
        "og_object": null,
        "comment_count": 0,
        "share_count": 1,
        "reaction_count": 1
    },

"Facebook": {
        "comment_plugin_count": 0,
        "total_count": 32,
        "og_object": null,
        "comment_count": 0,
        "share_count": 22,
        "reaction_count": 10
    },

"Facebook": {
        "comment_plugin_count": 0,
        "total_count": 8,
        "og_object": null,
        "comment_count": 0,
        "share_count": 7,
        "reaction_count": 1
    },

Perhaps not that critical since it is just a way to show some social proof, but I was expecting to have the numbers make some sense anyway and at least show a consistent result.

Is this a know phenomenon? Anyone tested this and found up and downward varying outcomes on the same url like this?

Thanks, Hans