A little sciencerant to pass the time

I was having an..well argumoid...the other day online with the delightful Rich Boden about whether or not one needed to collaborate in order to save time learning techniques.  I think there are points on both sides, and also think we were talking at cross purposes.  But let me spell out where I am coming from by an example in my own field.
My thesis is that the literature as it exists today is often an insufficient document of techniques to allow reproducibility.  This may or may not be recognised by those recording the results.  Actually I get the impression this is something they do not care about.
My field these days is microfluidics.  But just to show this is not new there is an example I am aware of from my birth-field, organic chemistry, where someone recognised this problem and actually put it right.
in 1984 Henri Kagan discovered that sulphides can be oxidised asymmetrically to sulphoxides using a modified Sharpless reagent.  This was good news, but many people struggled to repeat his results and boy did they try.  So later on Kagan pubished a much more detailed procedure and explanation to help others along.  This one worked fine.  I used it myself at least once.
The thing is that the original had described what they had done, but only gave enough detail for someone who was used to the system to effectively repeat it. Insufficient dissemination of knowledge.
In my field now this is a huge problem.  Microfluidics, when I started doing it, was a field only for a few labs.  partly this was because the fabrication facilities needed were expensive.  then the inimitable George Whitesides published a paper on soft lithography which showed you could make easy biocompatible devices from polydimeethylsiloxane (PDMS) quickly and do it in a small clean cupboard.  This led to a democratisation of the process.  When biologists wanted to do microffluidics they got a grant, then hired a biologist because microfluidics is easy and the lit teaches it to you, right?
Oh boy are they often in for a surprise.
PDMS itself is not so much a compound as a loose confederation of warring tribes.  For a start actual pdms is dimethicone, mainly a sexual lubricant.  It is a messy compound that operators of mass spectrometers hate because it gets everywhere.  Even in your lipstick.  But the PDMS used in chips isnt PDMS it is mainly Sylgard 184.  This compound is marvellous for many things.  It was developed as a diggable potting compound.  It is closely related to bathroom sealant.  But it was never ever meant to have chips made from it.
What is not mentioned much in the lit is how variable the results from chip to chip are with PDMS.  The pros know things like DO NOT MIX KITS (each kit contains monomer and setting agent, each is titrated.  the setting agent from one kit does not work well with another kits monomer) and TIME STAMP THE CHIPS ( PDMS continues setting and changing at room temp for weeks.  experiments on chips from the same batch at different times will vary).  We know the chips drink oil, leak water, transpire gas.  We know they will absorb dyes (put an m and m one and leave it...).  We even once did an experiment which showed, reproducibly, that the speed and DIRECTION of stirring of the premix could give variable results.
So you need to standardise a lot.  the big fellas know this.
But the democratised user does not.  I commonly hear tales at bio conerences of people who could not get microfluidic experiments to work and gave up after months of trial.  Also commonly I can tell them in 20 mins what went wrong and how to fix it.  So why dont we publish?
Try it.  We recently managed to get a paper published on the problems in a related field, droplets, which showed many of the problems and systematised them.  This took several rejections.  why?  Not novel.
Well no, not novel, just desperately needed.
And the rejections took a pattern.  professors rejected.  Students and postdocs who saw it at conference wrote to us begging for a transcript.  We distributed over 500 copies of our poster.  Why is this?
The people doing the experiments know the problems.  The professor only sees the good results at the end of the week.
This is a multivariate problem.  As scientists we edit our results, showing the working system.  the journals kinda want this.  But we run the risk of editing out vital information that allows reproducibility.  the number of microfluidic designs you only see once is too high.  And i do not think my field is alone here.