> If you are trying to trick somebody with your data, never share the raw data; only share the conclusions. Bonus points if you can't share the data because it is somehow priviledged or a trade secret. The last thing you want is to allow other people to evaluate your analysis and possibly challenge your conclusions.
Of course, I'm not against sharing data. However, the satire here is slightly too optimistic that people, when given raw data, will attempt to verify it for themselves. When people are given plain narrative text, they can still be profoundly influenced by a skewed headline -- something which everyone here may not ever be familiar with :)
I guess I'm being curmudgeonly about this...We should all share the data behind our conclusions, but don't think that by doing so -- and an absence of outside observations -- that you were correct. Most people just don't have time to read beyond the summary, nevermind work with data.
Well it is similar to open source. Just making a project open source doesn't automatically reduce the number of bugs. People need to be interested and capable of auditing it. However, even if only a small percentage of people do audit it, the gains from that apply to everybody.
I think it generally works out. More popular open source projects naturally have more eye-balls, which means more people auditing it. Similarly with sharing data, more controversial or interesting claims will naturally attract more attention and therefore more people who are capable and interested in verifying the claims.
Of course, I'm not against sharing data. However, the satire here is slightly too optimistic that people, when given raw data, will attempt to verify it for themselves. When people are given plain narrative text, they can still be profoundly influenced by a skewed headline -- something which everyone here may not ever be familiar with :)
I guess I'm being curmudgeonly about this...We should all share the data behind our conclusions, but don't think that by doing so -- and an absence of outside observations -- that you were correct. Most people just don't have time to read beyond the summary, nevermind work with data.