29 Sep 2004 | Justin French said...

Well put. It's one of those things I have in the back of my mind, but you've managed to bring up front and centre with a few concise paragraphs. Definitely one to add to my "personal mantras" :)

29 Sep 2004 | indi said...

Amen to the rant. In my job where I have to provide metrics that eventually get flowed up to our government customer I am sometimes flabbergasted by the nonsensical minutiae requested to be in the reports. Unfortunately I don't have the luxury of leaving out that data even though I know it's useless (and believe me, I've tried!)

29 Sep 2004 | Darren Delaye said...

Totals can obviously sometimes be at least as important as the individual data, such as when the total at the bottom is an amount of money.

If tabular data is separated into pages (the way forum threads are paged), one question I've run up against is whether the user may be confused about whether a total is the total for the page of items displayed or for all items on all the pages. The potential confusion is there if the totals are positioned as the last row of the table, even if the totals are bold or the row is in a different color.

29 Sep 2004 | Brad Hurley said...

However, sometimes statistics such as the number of tasks you've completed in a year or the number of e-mails you've sent contribute to a sense of accomplishment. They don't necessarily change your behavior, but they might affect the way you feel about what you're doing. I wouldn't change the way I do my website if I knew I had 12,000 visitors a month instead of 7,000, but I'd probably feel more inspired by the larger number.

What really bugs me about statistics that are typically displayed in interfaces is unnecessary precision. Stats can often be rounded to the nearest 10 (or in the case of large numbers, the nearest 100) without sacrificing the utility of the information; for example it's silly to say that the average visitor to your site downloaded 4.25 pages.

29 Sep 2004 | Paperhead said...

This fails to take account of nerdy interest though:

Would knowing how many emails I sent this year change my behaviour in respect to sending emails?

No.

Does that mean I wouldn't be interested to know how many emails I sent this year?

No.

BTW, Jason, there's a big flaw in your post:

That middle paragraph asks whether someone would change their behaviour based on a difference in the amount of email they send. What you actually needed to ask was whether or not someone would change their behaviour depending on knowing how many emails they had sent compared to not knowing how many emails they sent.

I tried to answer the question as it should have been asked.

29 Sep 2004 | Don Schenck said...

Reminds me of the old story: A guy runs a 500-page report on green bar paper and takes it to his boss. The boss says "I only want a summary".

So the guy goes back to his office, tears off the last three pages, and gives that to the boss.

We should be trying to prevent "information overload", not contributing to it.

29 Sep 2004 | Arne Gleason said...

Good rant. It seems to me that no matter how overloaded a design is, there always someone insisting that its absolutely essential to add more. Maybe because credit is given for additions but none is given for leaving things out.

29 Sep 2004 | Beau Hartshorne said...

It depends. I think you should leave that decision to the user. A title attribute in an <a> tag or an alt attribute in an <img> tag could provide this information without killing a clean design. If different regions of an image are significant, you could use a client-side image map to paint hotspots onto the image.

Edward Tufte's sparklines (large download) may be a good way to clean up detailed information.

Users have asked for detail like this in my software, but they do not want it cluttering up their interface. I think the real challenge is leaving the information accessible while keeping it out of the way.

29 Sep 2004 | DC said...

Removing something is a fairly primitive form of information design. The problem isn't information overload, it is the inability to make the judgement suggested in the kickoff to this thread: "Would knowing or not knowing this specific number alter your behavior in any way?"

What if we apply this concept to one of the infamous web site hit counter. If we ask this question, the visit counter becomes "Average orders take 2 minutes 37 seconds." Or any one of a hundred other ideas to change user behavior. Yes, but does it work? Home Shopping Network would probably argue it does. Geico uses something like this in TV ads, just the average savings of $189. Not fancy. In fact no code necessary at all. (and maybe that is the answer -- there's no geeky challenge).

The problem is not statistics on the web interface, it's not asking whether the statistic matters to change behavior.

29 Sep 2004 | JF said...

BTW, Jason, there's a big flaw in your post...What you actually needed to ask was whether or not someone would change their behaviour depending on knowing how many emails they had sent compared to not knowing how many emails they sent.

I think it's the same question. Do you /really/ need to know this information or not? Does knowing this information /really/ influence your behavior? Sometimes it does, sometimes it doesn't. And when it doesn't, leave it off the page. I'm not saying leave all totals off the page, just leave some off because some (and often most) don't really inform decisions or actions. My point is to think, not to do by default.

29 Sep 2004 | DC said...

My point is to think, not to do by default.

This is an important point to make, as it is far, far, easier to not think. Witness the lack of variation or alternative to web site visit counters. This could be an easy element to count any sort or kind of important business factor. Heck, they could put up service factors -- useful or not -- like how many minutes it takes a business to return a phone call. This turns "service" from empty filler content into information.

...it doesn't happen. Instead "web counters are bad design." Period. End of discussion. End of thought. Again, it isn't the counter that's -- it's the lack of thinking behind them.

29 Sep 2004 | Daniel Drucker said...

This is similar to what's taught (I said taught, not practiced) in medicine -- don't give a test unless the outcome of the test will change your treatment.

29 Sep 2004 | Paul said...

If I knew I had only completed 15 tasks this year I would be scandalized. So please leave that number out. It would interfere with my bliss.

30 Sep 2004 | Jonathan Lipps said...

Two things I have decided after reading the discussion:

1. Stats of any kind, detailing dubious relations and with values in custom units, are to be lovingly dreamed about, created with XSLT, hugged, and generally pampered and adored, by me.

2. Since about 0% of the rest of the world is as neurotically statistics-grubbing, stats of any kind should be left out of all applications whenever possible.

30 Sep 2004 | Paperhead said...

Hey Jason,

No, it's not the same question is it.

The overall thrust of the post is about an absolute, displaying data on a page or not displaying data on a page.

That middle paragraph is asking us whether our behaviour would change dependent on differences between two different sets of data.

See the difference there?

***

As it goes, my personal preference is for the data to be available, but not necessarily displayed. I think this also cuts to another problem with the idea as stated in your original post:

If you're sitting there wondering whether or not this data needs to be displayed, then you are one stage on from where you should have been asking the question. If your aim is for efficient programming you should have been asking this question further back up the tree, you know, before you actually gathered the data.

30 Sep 2004 | DC said...

Or, how would your behavior change knowing that youve completed 612 tasks this year instead of 429 or even 15?

It wouldn't. If the tasks were able to be ranked in some way, then you could begin to see how there might be some basis for behavior change. Otherwise the numbers have no meaning. There could be 15 tasks which take ten times longer than last year's 150 tasks.

There is no frame of reference. Tasks could be pointless time-wasting activity. Some tasks may be in direct opposition to larger goals. Still others might be repetition of past bad patterns or outright mistakes.

30 Sep 2004 | RS said...

If you're sitting there wondering whether or not this data needs to be displayed, then you are one stage on from where you should have been asking the question. If your aim is for efficient programming you should have been asking this question further back up the tree, you know, before you actually gathered the data.

Whether or not the data has already been gathered, we often need to decide which data will be useful to gather in the first place.

Either way, we're talking about conversations that begin like this:

"Wouldn't it be nice to show the last login time?"

"You know, we could show the total number of posts in that sidebar."

"What if we put the number of times the attachment has been clicked next to the file link?"

Sometimes the data is available, sometimes it isn't. But it is always important to ask: Will adding this data to my design help someone do something better? For me, that's what this post is about.

30 Sep 2004 | Paperhead said...

That first section just pretty much reiterates what I'd said in my previous post Ryan.

If there isn't a routine to collect the data, then the data isn't available. Agreed?

So make the decision as to whether the data is going to help someone do something better at that stage. If it's not going to be helpful, then there's a pretty good chance that you don't need to bother collating it in the first place.

If you're making a decision that you're not going to use some information after you've collected it, then you're making that decision one stage too late.

There, that's not so hard.

29 Oct 2004 | cheap cigarettes said...

Hi, if you want to find the best cheap cigarettes from all the discount cigarettes stores on the internet, enter our cigarette store and get yourself a cheap cigarette right now! We offer marlboro and all your other favorite brands. Thanks.