Do you use Stripe's risk evaluation API? In my experience, a lot of chargebacks come from people who have had their credit card info stolen, then file a chargeback when they see the unknown transactions in their statement.
You could also try using the two step auth and capture method to allow more time for the victim to realize their info was stolen and do an additional fraud check with potentially more up-to-date models right before performing the capture.
I haven't used CreditKarma's tax product either so I can't speak to its accuracy, but from my understanding they didn't build their own product, they acquired a company called AFJC which has been in the online tax space for a while.
Credit Karma missed some things in my NJ state return that would have cost me a couple hundred dollars. There didn't appear to be any way for me to manually include a correction, so I ended up filing with TurboTax (free version).
This wasn't even a particularly esoteric exemption, I expect it would have come up with anyone who worked for two different companies in the same year. Aside from that, I'm a very standard case.
That quote reminded me of something my mentor told me when I was training to interview people in-person. He said if they're really struggling, change it up with an easy question for exactly the reason you gave, but also to get their confidence up so they don't start the next interview feeling bad.
That's certainly a good data point in the 'against' column, but it really only indicates that any current, if present, isn't likely to be caused by the normal circulation of water from filtration systems or similar. There are a number of other things that could possibly be causing a current.
For instance the current could be caused by the interaction between the pool and the motion of the swimmers in the pool. It could be caused by the overflow wave from 8 people entering the pool. It could be caused by uneven gutter heights leading to the pool draining at a faster rate on one side/corner. It could be a combination of all of these things, none of which would show up in a static test like the one they demonstrated.
The fact that this has happened at some races and not others, and the ones it looks like it has happened at are Myrtha pools, is interesting in itself.
Those tests are with floats. That may be the standard but it isn't appropriate imho. Swimmers do not swim in the top few inches of water. They swim as deep as four feet in dives, and are constantly dipping hands feet down two or more feet. Any current down there would go undetected by such a float, especially when it is surrounded by lane ropes.
Do a test with something hanging below the float, below the protection of the lane ropes. Or drop some dye in the water.
(I'm a little surprise by the wave action visible in these tests. With so many ropes, that water should be like glass very quickly.)
There is some theatre here. Swimmers know the water moves, that it isn't a stationary tank. A perfect pool, with perfectly still water in every lane during an actual race, isn't possible. As slight winds can affect football games, swimmers just have to live with slight inconsistencies between lanes and pools.
On a more serious note this really is an awesome release. Kudos to everyone that works on Node.js!
Ever since the "reunification" of node.js/io.js there's been a whirlwind of great activity both in adding features as well as LTS releases (which are critical for long term adoption of the platform).
Sidenote - was just messing with Homebrew a few days ago, and found it a little bizarre that all of the recipes are in source control and require a pull to update, rather than being on a server somewhere.
It's a test they're running so not all users with ad blockers will see it. You can get around it by setting the forbes_ab cookie to false before clicking continue.
To add to your second point, it's also good to make people come to you if they need anything from someone on your team so you can filter out all the unnecessary interruptions and let them focus on their work.
You could also try using the two step auth and capture method to allow more time for the victim to realize their info was stolen and do an additional fraud check with potentially more up-to-date models right before performing the capture.