This the exact reason why I asked one of my banks for a physical token and refused to use their 2fa app. One less thing on my phone, I would have to travel internationally the next time I replace my phone, and the annoying little keychain can stay in a safe place most of the time.
While an Apex would me more convenient, I doubt that Iāll be able to convince said bank to allow me to use something that isnāt a digipass go 6 or their app. Still, Iām mostly keeping that account open because of the seniority and being able to provide a banking certificate that says that Iāve been a customer for a long time. Some citizenship and residency programs ask for that and I might still want to move in the future.
Now if I could convince the banks that I actually useā¦
Try clearing cache ā¦ try incognito / private to test ā¦ force refresh (CTRL F5, SHIFT F5, whatever apple does)ā¦ chances are you are pulling the old S3 bucket links vs the new CDN links
Using Firefox mainly, have tried incognito, Chrome, Safari on iOS, disabling pihole, safari on iOS on a different network (cellular vs home), and forced refreshā¦
Working on the magnet sensitivity tests. I made everything cleaner with neat little graphs.
Using a DB makes everything a lot easier.
This is the process for an Amplitude detection test , the most basic āhow sensitiveā question.
I switched to using a dumb binary search to find your amplitude threshold to a predefined precision for each tested frequency.
Itās faster than testing a range of amplitudes like I used to.
Also the āalgorithmā will now pick your next test for you, slowly filling your analytics as you go
Should I be using clearer names and notations?
Do things like amplitude detection and frequency discrimination make sense to everyone?
Also the graph currently shows the lowest detectable amplitude per frequency, so low values for more sensitive. So despite it being more transparent this way maybe I should invert it so that big values mean more sensitiveā¦
Just looking at your process, the graph and then hearing the tones, very much reminded me of a hearing test during a medical.
What if you had a ātestā mode, whereby it runs a random blind test at different amplitudes (no audio) and different cadence, and the user taps the screen when they feel the vibration, [if āYESā within 1sec = true]
On completion then graph and record.
Doing it this way means the user has zero reference therefore more accurate results.
It could be done Ascending, Descending or Random
Oh the audio is only there because Iām recording on pc. Itās not there when you use a pico
So yeah the tests are 100% blind.
You could intentionally cheat by knowing its a binary search but why would you do that
This is what I was doing before but it takes a LOT more questions to narrow it down than the binary search. And Iām trying to reduce as much as possible the number of questions because there are a LOT of them if you want to fully characterize your magnet .
Say you want to test for every 10Hz step on a range of 10 to 300. Gor 3 different signals. And each test take ~10 questions to narrow down on your score
Thatās 30 * 3 * 10 questions youād have to ask
And thatās just for amplitude detection. Other tests are incoming.
Right now Iām down to ~150 sets of ~6 questions to get a reasonably accurate curve for all 3 signals
On a 10-300hz range
Assuming a question takes you 2 sec, a set takes 12 sec. So you can do 15 of them in a day and be done in 10 days with this test
