Two Essential Features For Question & Answer Services


cc licensed ( BY NC ND ) flickr photo shared by ViaMoi

I’ve spent some time on Quora, Yahoo Answers, and similar sites, but the one thing that continues to frustrate me while I use them is that there are so many similar questions with similar answers, and there are so many answers delivered as attempts at humor or cleverness which do not really address the intent of the person asking the question, and therefore are clutter on the page.

These sites, however, seem sold on the idea of retaining everything within a question in some form. If you are trying to provide a service where people can get great answers to good questions, you need to be able to remove material that is not providing that value. I’m not talking about deleting controversial answers; those should remain available. But let the material of no substance go away. If you just gray it out and tell me it was downvoted, I don’t know whether it was controversial or just junk, so I still have to invest in clicking it and reading it. This is undesirable.

Secondly, and most importantly, there must be a way for two similar questions asked to be merged. Consider the following questions:

  1. How do I learn to play the guitar?
  2. What specific things should beginners focus on when learning guitar?
  3. What’s the first thing you should master when learning to play the guitar?

These are three actual individual questions on Quora right now. Quora and other QA sites become more difficult to navigate if these similar questions are not merged into one question. There is also the value added of increasing the likelihood of finding the one really great answer if you merge the questions.

7

An App For Making Sense Of Terms & Conditions


cc licensed ( BY NC SA ) flickr photo shared by Jason Michael

I’ve written previously about how to make a website or service’s terms and conditions (or Terms of Service, or EULA) not only more digestible, but also more enforceable. I don’t believe most Terms are fully enforceable if they create a significant burden to the user to process. I don’t believe they should be enforceable unless they are straightforward and clear, and can be digested in less than two minutes.

With that in mind, there is an opportunity for a group of brave individuals to distill these lengthy agreements from various sites and services and offer the abbreviated version to people via a website, app, or browser extension. This seems like a good crowdsourcing effort.

For example, if I’m getting ready to purchase a ticket from American Airlines, I have to agree to one of the longest T&C in existence. (It’s possible that if you stretched it out from end to end, it would be longer than the flight you would be taking.) So, when you found yourself faced with these terms, you would click a little bookmarklet in you browser (or launch the app or visit the site, &c.), and you would be presented with a much shorter version of the terms and conditions.

This clearer version would get rid of all the superfluous legal jargon about severability, indemnification, and jurisdiction, and would instead focus on a few simple things: what information is being shared and how it will be treated, what both parties will do and will be prevented from doing, and any other part deemed noteworthy or relevant to the end user, e.g., “by using the iTunes service, you hereby agree to name your first-born child Siri.”

In short, the app would boil it down to what you wanted and needed to know about the agreement you were entering into, and it would limit it to what you could reasonably digest in a minute or two. It might even give you the option to leave feedback to the site or service provider in question, taking them to task for any overreaching.

7

[11 Sept 2012: Here’s a move in this direction]

The Hidden Information In Amazon Reviews


cc licensed ( BY NC ) flickr photo shared by sneeu

While I’m OK with reviews going to a five-star system by default, it is becoming clear that reviews are largely skewed toward being overly positive. Because of this, when we look at book reviews on Amazon.com, you’ll see that the majority, it would seem, are rated four out of five stars or above.

Objectively speaking, that one star may be more personal preference than anything else, so most books essentially have the same rating at a glance.

But, I noticed that while two different books might each be rated four stars, the bar chart on the product page of each book showing the number of reviews received for each star rating could look significantly different. I started to wonder if there might be extra information about the quality of a book in the characteristics of the reviews bar graph.

I think there may be signature qualities to the graph that can indicate whether a book is truly exceptional or so-so. First of all, we have to have a large enough sample, say 100 reviews posted before applying this heuristic, and of course, the accuracy of this should increase as we have more reviews (perhaps you could couple the data with bn.com reviews, &c.)

Secondly, you’d want to look at the slope of the curve of the line that would be created if we made the bar chart into a line graph—-and this slope should look like an exponential curve (the exact equation TBD.) Specifically, look at the slope between the first point (five-star review number) and the second point (four-star review number). I would hypothesize that there is a sweet spot where the slope is not too great (where you have a cult-like following of devoted readers and reviewers) and not too small (where there is significant hesitancy to give it five stars, i.e., there is something lacking.) Here are two screenshots showing the difference produced by these effects (and both of these books have four star ratings.)

New-earth
Tipping-point

Now compare those two to the appearance of this graph:

48-laws

One particular important characteristic I believe is the relationship of one-star ratings to two and three-star ratings. A great book will have one star ratings from someone, but most often (with notable exceptions in politics and religion) the graph will show more four-star ratings than three-star, more three than two-star. But, the relationship between two and one-star ratings is a bit more complex. If there is a sharp upturn in the number of one star ratings versus two and three-star ratings, this might be a cautionary characteristic, but it may be the case that a small upturn is actually a good thing (as a great book is more likely to be attacked.) As mentioned earlier though, you need to control for certain subject matter.

Other characteristics could include the ration of three-star to five-star reviews, for three-star ratings are often the most objective and critical. So, a book with a lot of five-star ratings and a high number of three-star ratings relative to that number might temper the overly positive initial appearance.

In short, I believe you could offer a web application that would allow someone to paste an product link into the app and it would offer its analysis based on these refined criteria. It’s like reviewing the reviews, and you could do it for other products other than books, where each would require a bit of fine tuning.

7

Quora Meets Twitter?


cc licensed flickr photo shared by dkscully

I think it would be a simple matter to mark Twitter questions, that is, questions being directed to anyone and everyone, with the hashtag #?so people who would like to know what people are asking would be able to do so easily and respond to them. (We could just look for tweets ending in a question mark, but some would be rhetorical, and others would not be intended for exposure to the “hive.”)

It seems like an easy enough task to make a web page that would cull these hashtagged questions from Twitter and compilied them in one place in real time, as opposed to using Twitter’s limited search. As a matter of fact, I’ll suggest this to my Twitter developer friends, and I’ll update this post if and when they create something like this.

If it proves popular, maybe we’ll put onead on it, so as to pay for whatever hosting (or API calls) would be required.

7

A Snopes For Food Ingredients


cc licensed flickr photo shared by ecstaticist

Do you read food labels? I confess, I didn’t make a habit of it until a couple of years ago. Since then, I’ve discovered all sorts of interesting spices, flavors, chemicals and preservatives. Many of these are quite safe; some are even beneficial. Others, however, are a bit more questionable. Wouldn’t it be helpful to have information about food ingredients at the touch of a button without numerous Google searches? I hear a rhetorical, “Yes!”

Enter ingredients.com, or since that’s taken, some less direct way of saying it. It would be a website devoted to the latest objective information about food ingredients. What consensus there is about each ingredient would be presented first, followed by differences of opinion and the evidence for each. A score would be given to each, with 100 being the highest, i.e. the healthiest and safest.

But, here’s the great part. You’d have an iOS and Android app that would allow users to scan the bar codes of items at the grocery store, and each item would have a combined score, or perhaps color code. You could even apply it to restaurant menu items if you crowd source that information. You could then expand the site to items found in a drugstore, e.g. mouthwashes and creams.

It would need to be as objective as possible, and have a team of people who represent views from across the spectrum, from scientists to government regulators to ingredient manufacturers to doctors (both Western, Eastern, and alternative.)

7

The Beautiful Internet

This one is the graphic designer in me speaking out.

Let’s face it. A lot of the internet is ugly. Not metaphorically ugly, but from a design perspective, literally ugly. Whenever I run across a poorly designed website, I usually judge the book by its cover, and automatically assume the content, product, or service is inferior to one offered on a better looking website. For a long time, I continued to use Yahoo instead of Google because Yahoo looked better, cleaner, and more professional.

Of course, it is not always true that better design provides the best service, e.g., Craigslist; nevertheless, it is an impulse that often serves me well, and I’m not too interested in spending a lot of time challenging the assumption. I think there are a lot of other people like me, especially with regard to that last point. If it doesn’t look right, it doesn’t feel right. And if it doesn’t feel right, people move on.


cc licensed flickr photo shared by gualtiero

Well, what if we could separate the ugly internet from the beautiful internet? What if there was a search engine that had a design standard for the sites it listed? Would this offer some encouragement for sites (especially popular sites) to step up their design a bit?

Now, I don’t think everything has to be cutting-edge. There are websites devoted to showcasing the best and fanciest web site designs from around the web. This would not be that. Instead, it would set a bar that could easily be met with a minimum of effort.

Now, the interesting idea within the idea is trying to automate the aesthetic review process. I think if the offer were made, hundreds if not thousands of graphic designers would be willing to spend an hour or two a week just clicking YES or NO when shown various web sites. (They may need to pass a little visual test to demonstrate their familiarity with the basics of good design.) Each site would be shown to X number of designers, the results tallked, and if skewed significantly in one direction or the other would be considered reviewed, and where there was less of a consensus, the sites would be internally reviewed.

I think it would only take about 15 seconds to evaluate a site (on average.) That’s about 500 sites reviewed in the two hours our volunteer works every week. With just 5,000 volunteers (which I think is hardly a challenge), that’s 10 MILLION sites reviewed every month. While there are nearly 1 trillion unique URL’s, there are only about 300 million actual web sites. So, assuming zero growth, this would be a three year project.

Or you could write a program that recognizes good design …

7