Figure out what you really want to do with this. Do you want to understand everything very broadly? Do you want to become a specialist in a particular niche?

If the answer is broad understanding, lookup dogvomit's post and take the traditional coursework in the traditional order at your pace. There are sets of problems honed over the last 100 years to train people to think like physicists. Then you can go read Einstein and dense particle physics books; that's a lot of fun but probably won't go anywhere. Very, very few physicists contribute generally any more.

If you want to be a specialist, find a very well defined project you could really dig in to and enjoy. Something like one particular measurement you think you could do on your own. Pick something recent that you like. It's all out there on Google Scholar. Fill in the general physics you need as you go, but you'll probably need more engineering, software (and money) than anything else. Above all, please do get in touch with the people who inspired your work!

If you're able to successfully repeat a set of observations, or just do something that looks at all like what some grad student did 5 years ago, you will make their year by sharing it with them. If you can do that even once, you will be well on your way to contributing meaningfully to their field.

To put the time commitment in perspective, this is the kind of thing a new "generalist" physicist will do for 2-3 years full time while learning their specialty. Unless you really like this and find more than 10 hours a week to do it, it could easily take you 10+ years. That's ok. I'd be thrilled to find out someone outside traditional physics replicated my results from 10 years ago.

I think the majority of the scientific publishing culture and industry is bad for science. That said, this is not a fair criticism. It's entirely reasonable to tell someone you expect to see more data in order to publish and to start a conversation among the editor, reviewers and PI as to what is necessary to prove a point. Research is not a perfect process and does not progress in an orderly, predictable manner. There are going to be typos and blind spots in any paper.

In this case, obviously Nature should not have published in the end. We can't know how that decision was reached unless we see all the correspondence between the editor, reviewers and PI. It would be much more useful to the scientific community to see how the PI managed to convince the reviewers to allow publication, rather than to debate what is really a standard rejection response.

Tissue culture is definitely a thing.

Does that mean it's the only way to gather that information? Would it be possible to develop alternate tools?

Look. Whether you like it or not, we are developing the measurement techniques and hardware that are going to make this antiquated approach to biology obsolete. Feel free to yell and scream while the field passes you by.

It's not fair at all to link opposition to gain in function research to an "anti-science" mindset. You should be ashamed that you're resorting to that argument.

This is something which is seriously debated in the pages of serious journals, at scientific conferences and by government program managers. To link valid concerns to an "anti-science" crowd is political bullshit maneuvering.

There is a very real and valid cost/benefit analysis to be done on pursuing this work. As biology catches up to the physical sciences in scope and function, you're going to deal with the same issues we have dealt with (I am a physicist). One of those lessons is that scientists don't get to decide the purpose of our work. It doesn't matter what you write in your paper, or what the program manager tells you the purpose of the work is. It doesn't matter WHY someone does the work, all that matters is WHAT the work is. It's extremely naÃve to think an abstract in a research paper can properly define the purpose of a piece of research.

There are experiments and research paths we do not follow because the intellectual benefit does not outweigh the very real possibilities for misuse. You asked how you expect people to validate these hypothesis without the work? Take a page from physical science and learn to use computer modeling and limited experimental work in lieu of full studies. Do some tool development. Don't just throw up your hands and insist this is the only way. It's not.

This will require a cultural change, and there will be lots of hand-wringing over whether new results are valid, but biology will be a more mature field for it.

I've worked for the government in a scientific job, with a lot of IT folks. It was probably the most relaxed atmosphere I've worked in. No expectation to dress better than business casual. No expectation to work overtime. No expectation to really get anything done.

It's that last one that's really the killer. If you're not focused on getting projects done, first and foremost, then you're not going to attract good people.

A good engineer isn't necessary when the jobs at a government office survive only by making the right political and budgetary statements at precisely the right times. With very few exceptions, technical success or failure just doesn't have much influence on your career in the government.

Lastly, 140 engineers will make no difference. The federal government is huge. The office I worked in was a backwater, nearly forgotten location. We had a staff of 5000 people, about half of them engineers and scientists. There are thousands of engineers in the government right now who would love to work on meaningful projects. It's not a lack of talent or manpower that keeps those projects from happening.

The answer is no.

TFA says this in about 10 pages, with all the gory details.

Can we try no to have clickbait headlines? TFA is a blog called "Ask Ethan" so it makes sense for the title to be a question. A more appropriate headline here would have been "Dark Matter and Dark Energy don't Impact the Big Bang."

Well, that's a terrible summary. At least they linked to the actual paper.

Good on Charlie for getting all this press out of the paper. This is continuation of work started when I worked in his lab (thin graphene transistors can be made with e-beam lithography, that gets you a bandgap and you can actually think about making a digital transistor, this paper has better measurements and better e-beam lithography - there now you don't have to read either of the papers).

It's not clear that any of this stuff will ever be used as actual digital logic. I think it's more likely to see commercialization as an analog transistor in a sensor (reason #1 - no e-beam litho required). Someone from Charlie's group will likely be part of making active graphene electronics work out. He's got former students or postdocs at Intel and IBM, and there are at least two of us with graphene based startup companies. So, we're working on making graphene electronics something other than an academic curiosity.

I think we'd all rather see a world where China competes with the west in science and technology.

I am a scientist and I complain a lot about corruption in funding, publishing, and public representation of science; but as a whole it's a very honest and productive enterprise. This is much better than competing to see who can maintain the lowest cost labor pool or the biggest weapons.

There's a concept in your post that doesn't quite come across as clearly as it should:

People who are very successful academic scientists are only publishing for a few years, because they're able to go get significantly better jobs outside of academia.

The 1% of folks who are publishing for 16 years strait are very good at getting grants and publishing papers, but have failed during that 16 years to do anything sufficiently interesting or important to distract them from the academic grind for even one year. Most of the great professors I know have spent at least a year starting a company, working for the government, launching a spacecraft or some other very useful, but non-publishable work.

With medical devices efficacy and safety are very closely linked. If you're providing a product that monitors blood glucose and you do a poor job of it, your customer makes incorrect medical decisions that are potentially life threatening. The closer an app gets to providing such "actionable" information, the more likely it is that it requires FDA approval.

That said, this "can't be overseen" thing is silly. The FDA doesn't have the resources to oversee ALL smartphone health apps, they don't want to, and they shouldn't. There's no debate there. If the next generation of phones include electrocardiogram electrodes or a sophisticated spectrometer, the FDA is going to regulate the health software using those tools. That's really the news coming out of that FDA statement.

There's a big assumption here that large factories are wirelessly networked...

That's not a great assumption.

Universities generally insist that all IP developed as part of a sponsored agreement is owned by the university (as opposed to the inventors or the funders - the two normal ways of doing things). This isn't the "classic" troll behavior, but it's not much better. It has the same result of depriving the actual inventors (small business, professors, grad students) of an opportunity to commercialize their work. It deprives the funders (US government, non profits, small and large business) of IP they should rightly own as well as discouraging people from working together.

Almost all of the research done by universities is done via such sponsored research agreements, not internal funding.

I suppose I was one of the early pioneers in this field, I didn't know it had a name. A few years ago we published a paper on attaching three different olfactory receptors to carbon nanotube transistors and exposing the resulting devices to a half dozen or so chemicals while monitoring the responses. We were trying to produce something which was more usable (i.e. real-time) than the electrochemical methods described in TFA (to be clear, TFA describes very good work, we just had a different approach).

I wouldn't say this is a field which is taking off. It is significantly difficult to combine proteins with electronics. There are very, very few people/research groups who have the combination of abilities and experience to make these devices and properly interpret the results. More often than not, researchers perform laboratory, one-off measurements they can understand, but have no relevance to modern electronics or systems usable outside of the lab they were built in. Another common issue is performing measurements you don't understand, coming to conclusions that are wrong and sending the field off in a useless direction. It is very, very difficult to both build a good experiment AND properly interpret the results. The physics/chemistry guys don't understand the biology and the biologists don't understand the physics/chemistry. It can take many years to just learn to talk to eachother and stop assuming that "standard" processes, assumptions and statistics are applicable. Getting funding for this stuff can be a challenge, because no one really has claimed this field and none of the funding agencies (in the US, at least) seem to understand it. There are a handful of senior academics who can do this stuff, and a growing number of mid-career guys like me, but we're still a very small group.

If people are interested in what's going on with this field, I would recommend looking up the work of Phil Collins at UC Irvine, Ethan Minot at Oregon State and Charlie Johnson at University of Pennsylvania. I'm sure there are other good groups out there, but I know those guys are good.

No, that's NOT mentioned in the actual study, just in the press release. Not sure why they're speculating about that at all.