Really. This article expresses perfectly everything that is wrong with Wolfram Alpha (which still can't tell me how much an elephant weighs, no matter how many ways I ask it). There are so many stories of users doing ridiculous things because of such hubristic user interfaces. Hang out with non-hackers long enough and you will clearly see how "they create an incomplete model of the giant electronic brain in their own, non-giant, non-electronic brains."
Also, the phrase "non-solution to a non-problem" is the best thing I've read all week.
"weight of a" always seems to trigger the weight query. The problem is that it doesn't have general data for "elephant" and doesn't know to link it to "indian elephant."
It would be a lot easier if you could just browse the data. They could really have just used a good online retailer "drill-down" style interface ala newegg to make this easier.
Maybe the tool selection could happen after the initial query? The results from the default tool guessed by WA, then choices of other possible matches. The key word "weight" could suggest the weight query, then key word match "elephant" to give choices like Indian elephant, African elephant, etc. and allow the user to click which one she had in mind. Maybe some keyword matches for elephant from other databases listed after that.
The problem, though, is they don't have the amount of data Google can use to help predict what a user most likely wants from a given query.
I wonder if that would that then turn it into a "non-solution to a problem"?
The part about "non-solution to a non-problem" refers how the author thought that not only is the interface a non-solution, but that having datasets and visualizations would be useless to lay people (a non-problem).
I was asking the question whether it's really a non-problem. I figured that a nutritional label is a bad example, and that if WA had sports stats or product comparison data, the lay person might be very interesting, hence upgrading WA to a "non-solution to a problem"
Casting it as a "false affordance" was well done as well.
I had great expectations when I heard from Wolfram Alpha. The things Mathematica is able to do backed by a giant structured data store seemed mind blowing.
As the author said the visualization tools and the (still small) dataset _is_ impressive. I thought: We have all this data, all this visualization options and now we can create aggregate and munch all the facts of the world with an ad-hoc query language. But this simply does not work. There is hardly any functionality to really process data (aside from the predefined ways). And if you come to think about it: One cannot think of a way more complex transformations could be expressed with WA's natural language interface.
For me a better interface would look like this: The basic interface is some kind of full-featured expression language to navigate the data hierarchy (perhaps similar to SQL). You could then build some GUI to interactively navigate/parametrize data/transformations to build more complex expressions. To make the system more easily accessible you could then - and only then - add the natural language recognition system on top of this, which tries to guess some expressions from your input string and gives you some suggestions to start from.
That said, WA's interface leaves much to be desired. With Google (and its ilk), I can enter almost anything in, and get results. If they're not exactly what I want, I can refine it, little by little. It's like a gradient search in a sense.
With WA, I can't enter almost anything in, and get some sort of results where I can continuously refine my search. Instead, I get deadends where it has no idea what I'm talking about, and I don't know what else to do to help it understand.
In fact, it reminds me of playing the old text-based adventure games, where you have to guess what you can do in the "dingy old cabin with a door to the north." You go crazy asking the computer to "pick up mirror" and it replies, "thoust cannot pickth up the mirror" You really have no idea what you can do, and no hint out of the myriad of possibiliies. You end up playing a guessing game, instead of an adventure game.
I think google squared has the right idea. You can enter in some, and it'll spit back some sort of results, usually with columns you don't want. Then you can proceed to refine those results, by adding columns you want and deleting ones you don't.
I prefer to phrase it as "If thousands of smart people before you have tried something, you need to know why they failed and have some reasonable reason to believe that your approach is better, or you will just be wasting your time."
I use this most often in the context of someone popping up and declaring that they wish to create a "totally visual" language. I don't want to stop that one-in-a-million guy who might make it work, but just blithely letting someone waste their time isn't very nice either. (That's where the whole "encouragement at all costs" ideology falls down; encouragement is not free and the costs are paid by the encouragee, not the encourager; think before you encourage somebody.) Usually I just see someone spout the same ideas that have been tried tens or hundreds of times before; the excited person should take the time to examine those efforts before continuing on, because the easy stuff has been tried and quite a bit of the hard stuff has been too. This goes for many things.
I promise I'm not trolling, but that sounds a lot like Steve Jobs.
The only way to save wolfram alpha is to improve it. It is currently not a marketable product to the massive audience it is intended.
Aside from the alpha discussion, an additional useful point is the concept of too-intelligent interfaces. This is related to two other observations. One is that for high-throughput data entry (also programming) a nice GUI is really not the thing you want. You want to be able to navigate entirely with the keyboard.
The other is wsywig document processors. Serious documentation (such as that for a fighter aircraft, which when printed out weighs more than the aircraft itself, of for documentation required by the FDA for new drugs) is not really done with wsywig editors, but markup editors of different kinds. If you document is 300,000 pages, you want the pagination to be done in batch.
A memorable quote.
On a related note, there was an interesting C-SPAN BookTv program recently where the author talked about the revolutionary standardization of international freight shipping containers:
If you could use WA as a Mathematica console BUT with access to great built-in (crawled!) data and visualization tools, then it would be useful.
If only this were written the week before we were deluged with Wolfram Alpha articles and the attendant hype, here and everywhere else.
On to Chrome OS, then Mencius. We need another reality check on the latest planet-changer.
It's the same reality check as always: the majority of people don't understand it (and implications) well enough to really use it.
The computer revolution hasn't happened yet. It's underway, and it's going to take time to overcome cultural inertia.