Concrete is testing positive because they are looking for proteins, not a virus.
No, PCR itself is a manufacturing technology. It does not "look for something", it tries to make copies using template and tries hard. To do a PCR you have to provide a template and nucleotides to build up copies using template. If you already have a little amount of desired product, process starts immidiately and you get large amount of copies in low number of cycles. If you have no desired product seed to start manufacturing, and do cycles, nucleotides around still randomly could match the template. You have billions of templates, so earlier or later that random templating will give you a product that perfectly match everything that immidiately become seed and normal process start. So, if you will repeat process long enough without seed, you will eventually get something that looks like what you want.
The same, but in IT language.
PCR is a simple thing that have two templates made from sequences of 10-30 two-bit elements long (20-60 bits or 3-7 bytes) and random two-bits elements around. Let the latter be endless source of random file, like /dev/random. Also there is different sequences around, much longer than template, 10k-3G elements long. Say, some files. Algorithm continuously check all files (including /dev/random) for that two sequences of 10-30 elements and if that two sequences match, it make a copy of what between that two sequences including that sequences. If you have a desired file templates targeted on, you will quickly get a lot of copies of that file. PCR "test" uses third short sequence and if sequence match to the something in copy it marks the result file with prefix "positive". If you get a lot of files with positive prefix, than, they think that they found target file. However, that happens on billions of computers connected into network simultaneously and if any computer make a copy it shrares it with others. If there are no any seed file matching the template anywhere, after some cycles, with high probability some of that billions easily will find that short silly 10-30 elements (actually 20-60 bits or just 3-7 bytes) matches from /dev/random (or, may be find them in some other file you don't need to copy). And all network shortly will produce a copies of that random garbage, that have nothing in common with desired product. If it is a PCR "test", one of the lines of that copied grabage will probably have third short sequence to mark copies "positive".
So, if you run system long enough it will eventually began to manufacture copies of some random shit even if there no any seed at all.
That is why Kary Mullis insisted that PCR is not a test. It could "find" anything in anything, even if there is nothing at all. If you have an example to copy - PCR will copy it perfectly and fast. But if you don't have it and run copying cycle again and again, PCR will eventually start to make copies of garbage thinking that it is exacltly what you need. To use PCR you have to be shure that you have example of desired sequence in it. If you don't know if you have that example, PCR is completely useless absurd generator of garbage, you can't know if it copies the correct thing or some garbage it occasionally find out of environment made of nucleotides.
But PCR could easily begin to replicate garbage in first cycles. Templates are too short and the number of "copiers" actively trying to copy something is enormous in usual tests, so probability of occasional match is relatively high. And that is without accounting a possibility that templates matches could also be in some other RNA/DNA sequences usual for tested samples. So, rt PCR is useless too as a test. It will always throw false-positives even in perfect conditions. More tests you do, more false-positives you have. Even with completely fictional templates you will always be able to create epidemic of completely fictional, non-existent virus just doing enough tests. Even if probability of template match in first cycles is only 1%, you just need to test 100 millions to get a quick rise to million of "cases" you could use to scare people.
And in the case of SARS-CoV-2 the things even worse - you can't even somehow calibrate it and estimate false-positives rate of the "test", because you have no calibration sample in the form of pure virus isolate.
No, PCR itself is a manufacturing technology. It does not "look for something", it tries to make copies using template and tries hard. To do a PCR you have to provide a template and nucleotides to build up copies using template. If you already have a little amount of desired product, process starts immidiately and you get large amount of copies in low number of cycles. If you have no desired product seed to start manufacturing, and do cycles, nucleotides around still randomly could match the template. You have billions of templates, so earlier or later that random templating will give you a product that perfectly match everything that immidiately become seed and normal process start. So, if you will repeat process long enough without seed, you will eventually get something that looks like what you want.
The same, but in IT language.
PCR is a simple thing that have two templates made from sequences of 10-30 two-bit elements long (20-60 bits or 3-7 bytes) and random two-bits elements around. Let the latter be endless source of random file, like /dev/random. Also there is different sequences around, much longer than template, 10k-3G elements long. Say, some files. Algorithm continuously check all files (including /dev/random) for that two sequences of 10-30 elements and if that two sequences match, it make a copy of what between that two sequences including that sequences. If you have a desired file templates targeted on, you will quickly get a lot of copies of that file. PCR "test" uses third short sequence and if sequence match to the something in copy it marks the result file with prefix "positive". If you get a lot of files with positive prefix, than, they think that they found target file. However, that happens on billions of computers connected into network simultaneously and if any computer make a copy it shrares it with others. If there are no any seed file matching the template anywhere, after some cycles, with high probability some of that billions easily will find that short silly 10-30 elements (actually 20-60 bits or just 3-7 bytes) matches from /dev/random (or, may be find them in some other file you don't need to copy). And all network shortly will produce a copies of that random garbage, that have nothing in common with desired product. If it is a PCR "test", one of the lines of that copied grabage will probably have third short sequence to mark copies "positive".
So, if you run system long enough it will eventually began to manufacture copies of some random shit even if there no any seed at all.
That is why Kary Mullis insisted that PCR is not a test. It could "find" anything in anything, even if there is nothing at all. If you have an example to copy - PCR will copy it perfectly and fast. But if you don't have it and run copying cycle again and again, PCR will eventually start to make copies of garbage thinking that it is exacltly what you need. To use PCR you have to be shure that you have example of desired sequence in it. If you don't know if you have that example, PCR is completely useless absurd generator of garbage, you can't know if it copies the correct thing or some garbage it occasionally find out of environment made of nucleotides.
But PCR could easily begin to replicate garbage in first cycles. Templates are too short and the number of "copiers" actively trying to copy something is enormous in usual tests, so probability of occasional match is relatively high. And that is without accounting a possibility that templates matches could also be in some other RNA/DNA sequences usual for tested samples. So, rt PCR is useless too as a test. It will always throw false-positives even in perfect conditions. More tests you do, more false-positives you have. Even with completely fictional templates you will always be able to create epidemic of completely fictional, non-existent virus just doing enough tests. Even if probability of template match in first cycles is only 1%, you just need to test 100 millions to get a quick rise to million of "cases" you could use to scare people.
And in the case of SARS-CoV-2 the things even worse - you can't even somehow calibrate it and estimate false-positives rate of the "test", because you have no calibration sample in the form of pure virus isolate.