A good example of how badly wrong COVID science can go was the discovery at the start of the pandemic that the most influential epidemiological model had lots of programming bugs in it that made its predictions wildly vary from run to run. The cause was a mix of thread safety bugs, out of bounds reads, floating point error accumulation, mixing up variables due to horrible coding style in a 15,000 line C file with hardly any comments and almost all variables being single letter/globally scoped etc. They'd never teamed up with computer scientists even though their simulations were big complex programs, their code had never been peer reviewed because scientists don't do that, and when programmers pointed out their bugs, they tried to claim scientific models don't need to be reproducible anyway.
https://dailysceptic.org/code-review-of-fergusons-model/
https://dailysceptic.org/how-replicable-is-the-imperial-coll...
There was no inquiry of knowledge or productive debate. There was a bunch of amateurs, out of their depth, desperately trying to cover up how inadequate their own research was - and mostly succeeding. The institutions turned out to have no interest in the question of whether their scientists can actually program the models they're proposing, and nothing happened to Ferguson or his team.
That's what science is actually like, in the 21st century. Look too closely and it falls apart.
That's exactly why I trust the science and try to shut-up in discussions on topics that I have no idea about (with varying results, I'm only a human).
I trust that people who spend their lifes doing research can and do have productive debate (to certain limits - they're also just humans).