Wrong. The issue is that publishing is considered sufficient.
It should be publish or die. How do you know they're doing anything if they don't publish? they could be reading slashdot all day for all you know otherwise.
But as is made clear here, simply publishing and getting it through peer review is clearly not good enough. We need to increase what they have to do to avoid this situation.
For example... maybe one scientist pays another scientist to reproduce his work.
Maybe you have big collections of graduate students that as part of their process of getting a degree get assigned some random papers submitted by scientists in their field and they have to reproduce the work.
Obviously this isn't always possible... but whenever it isn't possible that needs to be put as a giant red asterisk on the paper saying "this work has not been reproduced"...
Do that and you're not going to get as much fraud or laziness.
I think that there is a common misunderstanding about the function of a publication. First and foremost it is a progress report of the scientist. This creates a lot of published noise - no doubt - OTOH it creates something that can be measured. This is absolutely critical to keep the scientific circus running (in a positive way). There are different ways to measure quality (which journal, reading an abstract, reading an article, asking by email) and scientific progress/quality is somewhat orthogonal to the publishing process. If you want to be sure of something be sure you have your act together to judge publications. The system can be gamed but it is not a problem in itself.