Science can be compromised by openness and there needs to be a way around this without directly compromising freedom of expression. This idea was brought to a head for me by musing on the possibility of recovering neandertal DNA and the inevitable possibility of clonng them and the equally inevitable wailing and whining by religious leaders, "ethicists," and the sociopolitical left who won't oppose it so much as exploit the results with much public display. Somewhat the same happened with so-called "test tube babies," whose rights were worried over and whom religious authorities claimed would be "born without souls." No, they are normal individuals. Clones, once the technology is mastered, will be normal individuals even if the twins of persons older, even much older, than themselves and not "soulless" humanoids committing all manner of perversions. Since neandertals would likely appear sufficiently different from us as do blacks to whites and whites to blacks and either to yellows, certainly some social problems will be invented involving them, especially if they are generally less intelligent than humans.
Which brings me to my point, that science and research need to be carried on with a moderate amount of secrecy to keep it slightly beyond the reach of neurotics, professional malcontents, scandalmongers, and political agents. As I see it this will help protect unmistakeably beneficial advances such as in vitro fertilization and genetic engineering until they can be applied effecively.
The use of an artificial language for scientific papers, Esperanto or, better, lojban, would help. Latin was once the language used by Western scientists and scholars since it was known throughout Europe by educated persons. An artificial language (as opposed to a natural language which would be accessible to at least fairly sizeable populations) could serve the purpose of scientific privacy. Nobody would be prevented from learning the language outside scientific fields, but few people would take the trouble. There might be legislation in different countries for the purpose of protecting science from interference by the media and the aforementioned undesirables. Ultimately, all research must be required to be revealed openly.
There should be contracts but not external laws requiring a certain amount of secrecy, with violators losing funding and employment or pensions.
Again, there will not be total secrecy as in some spy agencies, and this wouldn't serve too well. No doubt much urban legend and folklore will then surround science -- but such is the case anyway.