Because it's so broad and vague. It could mean one, some or all of sixty things:
- how can we individually avoid committing bad science?
- how can academics avoid... peers committing bad science?
- ... their department committing bad science?
- ... their university committing bad science?
- ... fellow members of professional societies committing bad science?
- how can we advocate for doing some/all of the above?
Your question also mutates. You start by saying a) "junk journals and articles... pompous titles, pay-to-publish [etc.]" but then you say b) "by junk science I mean pseudo-science (i.e. work done with flawed or frivolous methodology), deliberately faked results, and generally very low quality research".
These are each very different things, some of which have no answers, some of which have clear well-known answers. There will always be bad or VLQ research. Peer-review can prevent people publishing e.g. bad physics in recognized physics journals, but it totally can't prevent people splintering off and forming a new field/journal/conference. Grant applications are another thing too, and each grant source has different criteria, levels of rigor, punishment for bad behavior (or lack of). (Peer-review is not perfect either: it's slow, political, fallible and ridden with cliques.)
"Deliberately faked results" constitutes academic fraud, but it's comparatively rare, and I submit to you that many papers simply have no results at all, which is again bad but in a different, passive, way, since it obscures scientific method - unless they're review papers, which they're generally not. There is no shortage of such papers, in most fields.
There are also authors who crank out 10+ papers on essentially the exact same finding - would you call that VLQ? Then again, given reject rates and unpredictable backlogs, who can blame them...?
"Pay-to-publish" is again bad in a different way, not per se, but because it sidesteps peer review, disclosure, replicability, standard use of terminology, which are again all cornerstones of scientific method, so it generally results in a tsunami of crap, and cliques who manufacture plausibility by citing each other.
So, you asked at least sixty different and contradictory questions. The tl;dr is clearly we can't prevent someone setting up the Abkhazia Open Institute of Antigravity and spewing out junk, but we can monitor and publicize any misdeeds, and try to limit them getting access to serious funding. But you knew that already. There is no blunt hammer to prevent the rest of the things you list. Wherever there are economic incentives, however slight, people will respond to them...