As a scientist who has submitted papers to peer review at various journals (and even had one rejected by my peers) I'm getting a kick out of these replies.

You are correct when you say that reviewers and editors are not going to duplicate an experiment. They won't. That's not their job. Their job is to review the study design for obvious flaws and determine if the data supports the conclusion. They also have other duties, such as making sure a paper fits within the journal's intended "scope" (otherwise you'd have a neurology paper in a molecular biology journal or something crazy).

So, where does the data go? Most journals (at least all the good ones, and all the good reviewers will, anyway) demand that any sequence data (DNA, RNA, proteins 'decoded') be deposited in a public, government database and the access numbers be printed in the paper. Therefore, anybody who has questions about that can quickly refer to the primary sequence data.

Also, many groups make gene knockouts (they "erase" a gene from an organism to see what it does) or add genes in. In many/most fields, these authors are required to either send these constructed organisms to a public repository or state that they will allow others free access to what they made so this can be carried out.

Usually this is enough. This way, if a person makes a claim other groups can access their strains & sequences to try to duplicate it on their own. I've seen more than enough papers in my field where somebody made a claim (usually its the bolder, more outrageous ones that are disproven the quickest), and it's generally not well accepted because nobody else can duplicate it. Then somebody gets their strains or looks at their sequence data and finds out the first group was wrong for some reason.  

Of course, this is different from people fraudulently lying or being malicious with their results. It is very difficult to prevent liars from getting work published, if they are good enough. However, the steps listed above usually mean the person will be detected eventually. When that happens, these guys come knocking. If the allegations are true that a scientist lied or forged data, then they get an article written about their shenanigans here (pdf of a sample newsletter, see page #8 and on --- you can see all the ones ever printed here) and at that point you might as well apply for a job as a garbageman because you are done in science forever.

Problem is, this takes time. Good science and detecting fraud (which is probably less than one .01 of a percent of all work). I've worked on projects that have taken half a decade before their first publication. This leads into the next problem and associated rant. Much of the problem stems from the fact that people want their shit and they want it now (this is a cool article, I think I may write it up for a diary or something). And the sciences are grossly underfunded. We have money for (insert the usual litany of my complaints) but we don't have money for space exploration or other science & engineering projects. If we're mortgaging our children and grandchildren's future we might as well spend the cash on something that will benefit them.

Never compromise.