You only need one false positive test result that ends up with a colonoscopy to understand how costly and uncomfortable these errors can be.

(I'm using an uncomfortable and illustrative example, but a colonoscopy is honestly pretty safe and boring as invasive clinical diagnostic procedures go)

Perhaps a better example than you intended -- recently colonoscopies are being considered more risky as a regular screening tool, because the procedure itself can result in harms, including (rarely) death. Not an expert here, but it seems they are still used in a fairly wide age group, but the recent trend is to be more cautious about their usage, as particularly for older patients they are more risky vs. the potential benefits. [1]

However I think this concern would be reduced if we had better first-line screening like the tests in TFA; the harms come from using a relatively-risky assay like colonoscopy for regular screening in healthy individuals, whereas if we had better noninvasive screening, the colonoscopy could be reserved for patients where there's a higher probability of a positive diagnosis.

[1]: https://jamanetwork.com/journals/jama/fullarticle/2529486

> somewhat reliable marker

For the USA...

Something around 1.8 million people are diagnosed with cancer per year. At a 50% false negative rate, all else being equal it would detect 900k of those.

And all else being equal, at a 0.5% false positive rate, if it were used as suggested it would incorrectly diagnose cancer in around 1.6 million of the population.

So around 60% of the people it says has cancer wouldn't have cancer. I guess it depends upon what you mean by "somewhat reliable".

You don't see how causing many people to worry and get invasive screening per person you have a true positive for is a worse outcome?