I find the user's earlier threads to be more interesting. Foone deconstructed a digital pregnancy test and found that it's actually just a <$1 pregnancy test strip in the shell, with light sensors that check whether it shows one or two stripes, and then display the result on a screen.
As I think Naomi Wu said, the paper tests are extremely accurate in the lab, the majority of errors come from failure to read the output correctly (misinterpretation or otherwise).
So adding a digital readout that produces unambiguous results justifies the price.
It's easier to have a group of professionals program it correctly once than it is to have someone who only uses these tests at most every 11 months (assuming success) try to read it themselves.
The issue isn't that it's hard to read. The issue is that humans are psychologically conditioned to believe whatever the machine says and not their own eyes.
My wife works in a lab, has a masters in biology and I'm decently scientifically literate and we had a hard time being confident that we were reading the paper strip correctly by ourselves.
Is it a faint line? How strong does it need to be?
Is that a line or is it a smudge?
Did we wait long enough?
Well these online instructions are for a previous model, are they still valid?
It's not easy. But having a calibrated machine tell you "pregnant" or "not pregnant" give a lot more confidence about scheduling a follow-up test with the doctor
I agree with this. It confirms that most "digital" devices like this (for example, scales) are just introducing greater inaccuracy whilst conveying a false sense of certainty.
What makes you say that? This is exactly the sort of scenario where punting interpretation off to a simple algorithm is probably more accurate than leaving it up to untrained humans.
I read a comment in a reddit thread and they said the "pro-tip" was to take a picture of the pregnancy test and up the contrast to be able to better interpret results.
So a less wasteful idea would be a companion app to help with that process vs. more electronics in a landfill. (Although obvious privacy implications exist).
That could happen with any two tests. That second test is really faint. At first I thought "Ok, so they're both negative, so what?"
Not to mention, false negatives are far more common than false positives. If you have a false positive, then there's typically some odd health concern with the person.
As the digital scales go I was in the market for industrial scale almost exactly an year ago. At the first glance the stated accuracy is simply function of ADC resolution and maximal measurable weight. On the second glance it is mostly about market segmentation and certifications, because for all manufacturers I've considered getting order of magnitude better stated accuracy seems to be a matter of software license and different cal certificate (and the SW license seems to only affect the error term reported by the scale, not accuracy of the actual measurement).
