The role of trust in science

Dec 06 2010 Published by under STS

Egon Willighagen just posted that Trust Has No Place in Science.  His point is that Antony Williams asked if/how much people trust various chemical databases and Egon answered that he doesn’t trust any of them, he verifies. Ok.

So back to the old standard, Mertonian norms. It is a norm of science to practice organized skepticism. (Merton argued that this wasn’t skepticism about everything, but specific to scientific ideas and statements – this specifically isn’t about religion or patriotism). Scientists don’t believe things just on someone’s word, they need evidence.

Right. But what form does that evidence take? Egon says he verifies everything. So I guess there’s no need for the database then? I mean, if you’re going to re-run all of the experiments that provide the data. Or even read through the methods sections of the journal articles carefully. But wait. How can you believe the methods section of journal articles. Articles have been retracted for things wrong in the methods section. Even if you read through the methods section, how do you interpret the results? How do you know it was the right method to use? Well, then you must do all of the experiments that lead up to developing that method. And then you better redesign the instruments. Oh, and make your own reagents or whatever.

At some point you have to evaluate things, but then move on. Philosophers will give you careful descriptions of it, but there’s the idea that you need some assumptions for every empirical test (Duhem-Quine).

Then there’s whole areas that you’re not expert in that you need co-authors to support you in. Are you going to check all of their work and second guess them?

There is some trust – skepticism – but trust.

(nah, I’m not out of my blogging funk and this isn’t up to my standards, but I need to at least try to blog to get back on the horse 🙁  )

update: spelled Antony's name wrong! sorry.

5 responses so far

  • "Egon says he verifies everything."

    I hope I did not say that (I didn't)... Surely, I do not check everything; that would be unrealistic. You are making a classical mistake. You assume that not trusting something, is distrusting that thing. That is not what I wrote.

    The real point is, that I do not have to trust the database. I can compare database (as Antony does), I can check internal database consistency. I can (unit) test my tools. I actually do these things, because the current state is not good enough for me. Databases have errors, software have limitations. But no one checks those.

    To me 'trust a database' equals 'I was too lazy to check my input'. Not trusting a database is *not* the same as distrusting a database. Neither are relevant. I use them, or I don't. And if I do use them, I take responsibility for what I do. I will not hide behind... etc, etc. That is the point, not that you must redo all experiments. That is not what I wrote.

    "There is some trust – skepticism – but trust."

    Yes, there is. But there should not be. There are arguments and facts. That's all that matters. Your paper writes "We started with the assumption that...", "It is know that..." ... you make your claim, and move on. No trust involved. I have yet to see the first paper that writes, "You know this chap from MIT? He wrote this nice story where he said that arsenic... His results look flaky, but I trust him."

    I trust I made my point clearer now.

  • OK, simple use case that just popped up. Say, someone did boiling point experiments, performed the experiment in triplicate, and reported the results in JACS.

    Do I have to trust this paper? No. I accept the results for true, and work with them. If I suspect other motives, then, and only then, I can redo the experiment. I do not have to trust the results to use them. If I use them, I take responsibility over those results: I state them I consider them true, and that I have no reason to think it is not.

  • Christina Pikas says:

    I think we differ in what we mean by trust. I believe you're defining trust as only blind trust - almost blind belief. Ugh - I hate that I'm going to quote this - but in Speed of Trust training that our department went through, we talked a lot about "smart trust." Not being a fool and just trusting everyone blindly or for poor reasons, but evaluating the statement/data, and extending trust only where warranted. This seems similar to some of your statements.

    I think that Antony could have chosen a better word than trust in his survey. "which of these have you evaluated and decided you could use? which of these would you prefer to use based on your evaluations of their merit?"
    edited to correct Antony's name

  • [...] Pikas: The role of trust in science Christina has a comment “I think that Anthony (sp.) could have chosen a better word than [...]