Is Stackexchange a good idea?

Question

I am only in my second week with this site and I have enjoyed many of the questions and answers and discussions but as I get to understand the purpose of the site and how it is being used, I have some concerns.

1. The site provides experts in all disciplines prepared to answer technical questions. The experts strive to provide the best possible answers. My experience with Cross Validated is that there are a lot of smart people with specialized expertise in R, classification/machine learning and many other important applied areas in statistics. This provides a great opportunity for someone who is stumped on a problem to get top-notch consulting advice for free. So I ask whether this is really a good thing since it could take away work people who do professional consulting for a living. Also are we encouraging laziness as someone who has a problem they find to be difficult might instead of thinking out the problem and doing their own research to find the solution decides it is much easier to raise the question here and just see all the great ideas poor out with practically no effort. We even help them reformulate poorly posed questions.

2. Some students come here for free tutoring. There seems to be an awareness here that this poses problems. Students should study the course material and try to solve the problems themselves. Cross Validated asks that these questions be labelled as homework and the approach to answering is to give guidance rather than a complete solution. Nevertheless doesn't this also also encourage laziness as the student can just pose it here or in several places on StackExchange to get as much help as possible before doing any of the work on their own?

Answer 1

Your notion of stealing business from consultants is predicated on the weird but commonly-held notion that the work is theirs to begin with. To me, that's akin to saying that people who have potluck dinners are stealing business from caterers. It's not their business in the first place but rather a task that is outsourced to them when it's convenient, and any caterer who blames potlucks for their lousy business probably has much greater problems.

The very first step into commercial food provision entails a very significant amount of work - finding a trustworthy company, agreeing on contract terms, payment method, coming up with the cash, conveying accurately what you want done, etc. That initial cost gets easier to swallow as the scale of the project grows, but for small questions it's absurd. Potlucks mostly just need a venue (conveniently provided for us by Atwood and Co.).

Potlucks also have all kinds of benefits that aren't really available in the market - a greater variety of dishes is possible (some good, some bad); the satisfaction of human connection, community, reputation; an opportunity for people who can't or won't be caterers (which can be for many reasons other than skill) to share their talents; cooking for the simple joy of it. I bet most professional cooks love potlucks - it provides an opportunity to experiment outside of work, get new ideas from others, cast aside the restrictions that constrain commercial cooking, and do something they enjoy in a low-pressure environment. Is there a risk for freeloading? Sure - but the community can regulate that without contracts and legal enforcement of obligations.

The bottom line is that this site provides something that consultants do not, which is low-friction, zero-cost access to a diverse body of self-validating experts who do this for the love of it, with great potential for learning, reciprocity, pride, community. Whether this site exists or not, that's a service that can't be bought.

Answer 2

To address various problems brought up by your question in turn, from my perspective:

"it could take away work people who do professional consulting for a living."

Frankly, if the services of a particular statistical consultant can be replaced by a few questions on CrossValidated or StackExchange, they're not providing enough service for their clients for me to feel bad. Beyond that, there are a considerable number of answers on CV that amount to "You're in over your head. Go find a statistician, use these words, and they'll be able to help."

"Also are we encouraging laziness as someone who has a problem they find to be difficult might instead of thinking out the problem and doing their own research to find the solution decides it is much easier to raise the question here and just see all the great ideas poor out with practically no effort. We even help them reformulate poorly posed questions."

It's possible, but if they don't put the effort in, they'll get little lasting knowledge from 'Give Me the Code' questions. For example, I on occasion post questions like that, but generally it's "I don't even know where to start with x..." and the question is meant to be something I can build off of and learn from. It's very hard to distinguish between that and a 'lazy' question.

Beyond that, being neither their boss, professor or mother, I don't care. Laziness as a vice is a particularly...puritanical notion. If someone just wants to get their plot, p-value or test name and go away, it's little enough to me. And I'd rather tolerate that than turn away someone genuinely floundering who could use a hand to produce a meaningful, robust and thoughtful analysis.

Some students come here for free tutoring. There seems to be an awareness here that this poses problems. Students should study the course material and try to solve the problems themselves. Cross Validated asks that these questions be labelled as homework and the approach to answering is to give guidance rather than a complete solution. Nevertheless doesn't this also also encourage laziness as the student can just pose it here or in several places on StackExchange to get as much help as possible before doing any of the work on their own?

Students also come here to ask questions when they don't have access to teaching resources, are trying to self-learn and have run aground, or are looking for a deeper understanding. I'd suggest the problem of a freeloading student is relatively minor compared to those things - especially as CV is fairly adept at not giving complete solutions.

Students being able to get "as much help as possible" while they learn isn't a bug. It's a feature.

Answer 3

One thing that comes to mind is that this site doesn't really provide "full service" consulting - we answer conceptual questions, and leave it to the poster to do the more time consuming work (e.g. implementation, writing up the results, making it all into a coherent story for a manuscript). If we were full service consultants, and commanding a large hourly rate, we'd likely be doing all of that work as well, which makes it sort of fundamentally different in my opinion.

Saying that we provide free tutoring reflects a pretty broad definition of tutoring - many of the homework questions presented here are ones many of us can answer at a glance, so simply asking leading questions (which is generally how homework is approached) requires very little effort on our part, and, in my opinion is not close to as involved as actually tutoring someone. In the case where the poster doesn't have the background to understand the leading questions, the question sort of "dies" - if we were tutoring, that would be the time where we'd have to step back and review more basic concepts.

Answer 4

I don't see any problem.

Any site where people can learn will be targeted by the lazy as well as the diligent, the homework-answer-seekers as well as the professional and hobbiest. Nothing can be done about that except community self-policing. Which seems, in my opinion, to be remarkably effective.

In terms of stealing business from consultants, none of the questions I've asked over time have any budget to hire a consultant to answer. I'm a student and hobbiest in statistics, and if there were any budget for what I do, it would be for university classes or a new computer, not a statistical consultant. (Unless there were an absolute statistical expert in my area who for some charitable reason would tutor me for a fee I could afford.)

I also do my best to answer questions that I have learned about or have experience in. (I'm an Artificial Intelligence guy by training.)

So my answer is "Yes".

Answer 5

Small chipping in: There were similar arguments as brought forth by @MichaelChernick when R started to gain momentum with version 1.0.0. I guess it's just a fundamental issue with community effort like open source software, open access journals, open data initiatives or community-driven open knowledge spaces as this one. Much has been written on what makes contributors tick and how commons influence for-profit competition, so I won't reiterate here.

But for me it is clear: A site like this serves the common good, one way or another. Reputation is the currency and fun and new insights the reward. That's why (most) of us probably do this.

Background: I am an outside professionalist and editor of an open source journal and R contributor and here on cross validated. I don't see my work threatened. If anything, I feel like I learn a lot here that will benefit my clients and my work.

Answer 6

Yes, I do not think this site can replace a consultant. It's more of pointing learners and other experts into the right direction and discussing with them. If you are concerned about career implications, the problem might be more that you help rivals for jobs and consultancy opportunities. But to compete with you with just the help of this site, they have to be already more or less in the same league. No company needing statistical consultancy will be able to just use this site instead, it might just help their resident expert.

As for laziness on the part of students, well, in my experience there are always ways anyway. They can write off from co-students or try to find the answers in textbook... If they post here either they are terribly lazy and people will notice and probably not give full answers (that's my experience) or they will format the question carefully, explain where they are stumped and get valuable advice from which they actually learn something.

Answer 7

Clients who have money find it more reasonable to just pay a statistician and be able to hold him/her accountable for the deliverable, and get it on the agreed time. The CV site provides a rough product, at best, that the OP may or may not use, at their own risk. One cannot always share the complete data set; the posts where the OPs would refer to the data on their webpages are not very frequent, so the answer will inevitably be incomplete.

On top of these reasons against CV being a threat for professional consulting, the OP may or may not be able to discern a good answer from a lousy one if they don't have good enough clue about it to begin with. Professional statisticians are likely deliver concordant answers... except in situations when there's an insurmountable paradigm difference between Bayesians, data miners and/or mainstreamers, for the lack of better terminology. Voting scheme is fundamentally stupid: 10 voices of a rep-1 person are not worth 1 vote of whuber or Michael Chernick, but the system does not allow whuber's vote to be worth 10 points (which has a reverse wisdom, in the sense that whuber still provides professional rather than populist answers). Bad answers rarely receive good scores, but sometimes not-so-great answers receive higher scores than answers that I consider to be a much better fit.

I remember there was a passing discussion of the mailing list of the Statistical Consulting Section of ASA, and there was no sense of a threat coming from this site.

In a way, the website works as an integrator/referer: a lot of my posts would be a reference to Wikipedia and an explanation, in my own words, how the Wikipedia material applies to the original question. It also works as a filter: the task of finding the right paper on arXiv would be absolutely daunting, but there may be somebody on the site who just read it, and can point to it. So it mostly saves the search time, rather than solves the problem fully; and most users are fine with that, and are able to take it from there.

Answer 8

I can only speak for myself, but I think stats.SE is a great idea for a third category of people you might not have considered as such: self-learners!

I am a Computer Scientist without any background in statistics, and I am learning lots on my own as part of my recent efforts to get into machine learning. I can hardly go to a consultant when I don't understand a description in a stats book, or when one of my toy algorithms isn't behaving as it should. I don't have a practical problem I need solving, just trying to understand some concept/algorithm. On the other hand, I'm also not a student who has a study group of peers and/or professors I might discuss my issues with.

What should someone like me do? Give up machine learning and all things statistics and leave it to the consultants? :-)

Answer 9

I strongly agree with @Erik and @Macro, but I wanted to point out something else that is intended to answer your question in the title of this question: "is Stackexchange a good idea?"

I think I have a privileged view of what you are intending to say because of my background. I'm not a statistician myself and my background is astrophysics (I only have a Bs. Cs. and if everything goes ok, I'll be in grad school next month). However, here in Chile the college curriculum is in some ways way behind those of the top universities: in my case, we don't even have basic statistic courses, so I've pretty much learned all by myself (reading books, papers, etc.). Think of my status: here in my area, only the professors with who I work have a working understanding of statistics. Everyone else thinks that statistics is some magic "trick-bag".

Because of all of the above, I think I'm pretty much on the middle. In one hand, it is true that I'm not a professional statistician, but I think I can help with some topics that I've studied for the work I do. On the other hand, I also have used this site to get some advice or simply when I get stuck on something. However, there is something nicer that I found here: since I discovered this site two months ago, my knowledge on probability and statistics have improved enormously. I think that's the beauty of this site: it provides not just question and answers, but train of thoughts. I think that's a little different from consultancy, where you just give out a report citing sources and trying to put in simple words to your clients something that you did that in fact requieres a profound understanding on probability and statistics. If you hire a consultant is simply because of (a) you don't have the neccesary background to work a problem by yourself (or not have the time to learn it) or (b) you just need results.

However, in this site that knowledge of the background is somewhat implicit (ok, very explicit in some cases), because you wouldn't get a lot of what is said here if you haven't studied or worked on the area. With this in mind you have two options when an answer to your problem pops out: (1) you try to understand an answer and, therefore, if you don't have the neccesary background you'll have to study it (which is not needed with a consultant around), (2) if you happen to understand what to do you can just take the reply and use it without knowing what you are doing (which is a common mistake in every area but this time, unlike with a consultant, you don't have a safe source to rely on).

As for your point with students, well, I think you can never really avoid lazyness. You could say the same with problems solved in textbooks. I've seen a lot of people that when they see a problem set, they inmediately search on the web for the solution to the problem and when they find it, they just copy the solution don't even understanding what they are doing. However, as I said, here you are unlikely to find solutions. You are more likely to find train of thoughts and from there get to the solutions by yourself.

So my answer is yes: it is a very good idea.

Answer 10

May rather be a comment, but there evolves a discussion here (which shouldn't happen anyways...) so my "answers" try to sort the different lines in the discussion:

But I see plenty of people enter the site who are very naive in statistics but have a project at work that they need help with.

But what is their alternative? Usually not to ask the fully-grown statistician next door - I guess most people would do that rather than typing in the questions if they had the possibility.

So the pro of this is that statistics knowledge is spead among people who do not have a statistician at hand. Also, sometimes people are told to go and find a statistician to consult.

About the naive questions: if you're coming from another field you may use a very different language than what is common in statistics. This may add a lot to the naive sound of a question.

Answer 11

Students should study the course material and try to solve the problems themselves.

That's notion is wrong. The point of university isn't to teach students to be able to solve problems by looking in a lecture script or their notes from the lecture.

University is supposed to teach students to learn something that will help them in their future lives. In today's world using the internet is a central part of learning.

When I do homework and have a problem with some strange R syntax and spend an half an hour trying to figure it out then it makes no sense to stay offline and reread "course material".

In a perfect world I should be able to solve my R synatx problem by running a Google search and finding some explanation of how that specific R function works.

If that approach doesn't work it makes sense to condense the syntax question to a question on stackexchange. It helps my with my problem. It however also makes the internet a better place. If another person has problems with the same syntax he can now find my question on stackexchange with answers to the problem.

Answer 12

People are using it, ergo it is a good idea.