Sometimes I have simple ideas that can be useful. They could be obvious or not, or it is possible that no one ever considered them the way I did.
For example:
To gather information about the statistics of the unemployed in a city or state, I suggest to build a website in which unemployed people register and enter their information. This information is very useful for decisionmakers on unemployment. But, after a while these statistics are not valid. On the other hand, unemployed people are very reluctant to update their employment status via the Internet or they don't have access to it. To solve this problem, we can send an SMS to them and they answer with 1 (as employed) and 0 (as still unemployed), then we integrate these answers to the central database. This way, we have up-to-date information in periods.
The idea was that simple, however maybe no one in our country implemented it. Could it be a paper? If yes, what can it include? Because it is as short as the above. Should I, for example, explain how we integrate the SMSs to the database (however, it may be simple too or the subject of other tools or papers). If I implement the system, should I provide the statistics of unemployment in a city or the percentage who contribute the plan? Totally, I don't know what else such a paper should cover.
For another example, suppose that I am the first one who invented the sliced bread. How long could my paper be and on what would I probably argue?
However it was just an example and after some research I may realize it is not workable, but I should say I actually built the website for an organization in my city one year ago (However not the SMS and updating part), then I was thinking if I can make a paper out of it. Maybe I could use some parts of the real data I gained in such paper, for example the people who registered (which was more than 80% of all the unemployed ranging from 20 to 40 years old) and those who had a cell phone (which was more than 98% of them), this topic shed some lights on my way.
Answer
I like papers about simple ideas. (I am writing one right now, hope others like it as well.) They are far easier to communicate and understand than complex ideas.
Then again, the question is why nobody else has thought about an idea if it is all that simple. In your specific example, the idea may not be workable, because people may simply delete the update SMS without replying to it. (And those that do answer may not be representative of your sample as a whole.)
So I would say that writing a paper about a simple idea is good, but it needs to meet the same conditions as any other paper: it needs to show that the idea actually works. An idea by itself is usually not worth an entire paper. Having the idea is often the easy part. Showing that it works is where the actual work happens.
So: Build your website for one city, let it run for six months, then write a paper about what you learned.
How to show that something "works" may well be the hard part. (For instance, in some parts of machine learning it is easy to "show" that a method is better than an established method by testing both on many datasets but then only reporting those on which the proposed method is superior.) Some journals/conferences/reviewers may be more stringent about what they consider proof that something "works". You may be able to get a publication out of a proof of concept by just building the website, without running it productively. Or by running it productively, but without assessing in some way whether the statistics collected by the website are actually more accurate than those collected in some other way. Look at what kinds of papers your target venue or community publishes, and let yourself be guided by that.
EDIT 2016-04-07, about that article based on a simple idea: it turns out that this simple idea (a randomized probability integral transform) was really good. So good, in fact, that multiple people had had the same idea previously, and at least partly independently of each other. Happily enough, a guru in the field pointed this out to me when I circulated a preprint and didn't savage me, but pointed out shortcomings of the rPIT and possible new lines of inquiry. The paper has just been published.
Bottom line: your simple idea may be good, but chances are that those ideas that are both simple and good have already been worked on.
No comments:
Post a Comment