They assumed the government's software would do this:
(1) Purchase #1 for $25 successful --> reduce remaining limit from $10000 to $9975.
(2) Purchase #2 for $10000 rejected and refund pending --> DO NOT reduce limit since you know the purchase didn't go through.
(3) Purchase #3 for $9975 within limit --> allow purchase, reduce limit to exactly $0.
Instead, at step 2, the software seems to handle the purchase by unconditionally taking $10000 off the limit at purchase time and putting $10000 back later when the refund is processed. (And consequently, at step 3, it failed the purchase.)
There isn't an obvious reason to do it this way, so it's surprising. It seems like if you can reduce the limit, you can check if you've exceeded it and just fail the purchase and never modify the limit (at purchase or refund time).
Oh, I think the email is ambiguous! I had a totally different interpretation, although I absolutely see yours now.
I interpreted "refund of the excess purchase" to mean that it is a refund of the whole purchase. If it were a partial refund, they could have just written "refund of the excess".
By adding "purchase", they made it sound to me like there are two categories of purchases (excess and non-excess) and that this purchase belongs to the excess category. That is, I saw "excess" as an adjective that says why the refund is happening, not how much is being refunded.
The more I analyze the wording, the less I'm sure which interpretation is right.
Me too! My interpretation was obvious to me at first, but after reading your comments and others I can easily see it the other way. I still think my interpretation is more natural, but there’s no question that they should have been less ambiguous. How easy it would have been to use a few more words for more clarity. Also, ideally, the actual amounts in question should have been spelled out in the email.
If I have a $10K credit limit, I charge $9975 and then I pay $9975, my credit limit is automatically increased back to $10K before the payment goes through.
She didn’t have a completely incorrect mental model. But the credit card company is being a lot more lenient than the government is being and the government should be more strict.
I expect the logic for allowing a transaction looks at all 'purchases', including the current one, and sees if the total is over 10,000. If it is, throw an error, send an automated email, and let a human deal with reversing the transaction when they see what happened in the error log.
(1) Purchase #1 for $25 successful --> reduce remaining limit from $10000 to $9975.
(2) Purchase #2 for $10000 rejected and refund pending --> DO NOT reduce limit since you know the purchase didn't go through.
(3) Purchase #3 for $9975 within limit --> allow purchase, reduce limit to exactly $0.
Instead, at step 2, the software seems to handle the purchase by unconditionally taking $10000 off the limit at purchase time and putting $10000 back later when the refund is processed. (And consequently, at step 3, it failed the purchase.)
There isn't an obvious reason to do it this way, so it's surprising. It seems like if you can reduce the limit, you can check if you've exceeded it and just fail the purchase and never modify the limit (at purchase or refund time).