Skip navigation.

User Centered

Studying the design of everyday things

Usability Design Errors that Cost Lives

,

This is speculation so I leave you to make your own judgment.

You may recall the Buffalo Crash last year that killed 50 people. One of those being a man in his house on the ground.

What's interesting here is they are starting to release their findings and while training and fatigue are critical. So to is the design of the safety system used for command-and-control of the plane. The system, in the event of an imminent stall (based on computer calculations, not sensed-data), shakes the control stick used to control the altitude of the plane. Then, if a stall does start to take place, the stick actually "fights" for control by pushing the stick forward.

As I read the following:

Originally posted by NYTimes:

The alarm consisted of the control column shaking. Aviation experts say this may have startled the captain because the plane, five miles from the runway, was already at low altitude.


It struck me as quite apparent that, "Oh, my, he thought they were crashing." Even just instinctively, a panicked sort of response. Perhaps he had time to revise this assessment and better training would have done that but the stick's "safety feature" is clearly what startled and panicked the pilot. Even if they reacted wrong.

Imagine it, your driving along, your car is in the wrong gear and all of a sudden it starts to shake the steering wheel. If you aren't fully alert, you might panic. You might even act on the impulse to grab the steering wheel tighter and pull it in closer to attempt to regain control of it.

Even more compelling is this:

Seventy-five percent of pilots who had experienced the stick-pusher activation in training also responded by pulling back instead of pushing forward, even though they knew ahead of time to expect a stall, investigators said.

(AP - 1)

Yes, 3 out of 4 pilots who knew it was coming did the wrong thing! (In this case, not doing anything would at least have allowed the stick pusher to get the plane out of a stall.)

This is truly astounding. Even 5% seems like too many. Somehow this safety feature causes a natural reaction that is the exact opposite of the action that should be performed. It would be okay if it merely caused people to be startled and confused *and* then figure out how to react. However, it seems clear in this instance it actually prompts an improper action.

The designers here do not appear to have tested their safety feature but trusted it's intrinsic value, "Warning for stalls is a great idea!" -- but we should take care to not have warnings that cause bad situations to get worse.

This is especially important as consumer cars and trucks begin getting advanced computer functions that could very well alert drivers to dangerous conditions. Just don't make my steering wheel shake.

Trickey form flow

, ,

I would design this login form to respect the "flow" of the fields with the eye. The "go" button is visually disassociated with the name/pwd fields, compared to the "why login on function"

I would swap 'em.

(if you follow the link, you just gotta check out Petey and Pokey! Best mascots EVER!)

Best way to ensure consistent results in your survey

, , , ...

Jeff Harrison found this screen while taking a survey for JC Penny's:

Future magazines and photobooks

, , , ...

A hybrid proposal of several very good user interactions.

Read more...

User Centered Design...HTC HD2- ringer shuts off when it's picked up!

, , , ...

It's a simple matter that just about any modern smartphone can do, but it's the fit & finish that make a great user centered experience.

The HTC HD2 (which I'm impressed with) uses its accelerometer to silence the ringer when the user picks up the device. Great for those that fumble around in their purse or computer bag because they forgot to switch to vibrate (use Locale on the Android platform!)

Opera Unite Review - Part 3

I apologize for the delay in this posting... While this fulfills an internal drive at bettering the world it doesn't quite pay for my rent. Job hunting bites.

In Part 1, I discussed what Opera Unite wants to be. In Part 2, I reviewed the current offerings and my personal use cases. In Part 3, I'd like to discuss why I feel the underlying architecture seems flawed from a usability perspective and how that demonstrated itself in Part 2.

As a reader pointed out to me. Opera Unite is not the Apps it runs, Opera Unite is the platform that runs it. However, I contend that is semantics, you cannot separate them. Without the Apps, Unite is nothing that can be used. You cannot use one without the other. In this symbiotic relationship my review is Opera Unite's usability as it appears is for the end-user, it is simply "Opera Unite" (which is how Opera's own marketing targets Unite to the public). Only the developers will be truly interacting directly with Unite in a way that has elements of usability. Besides, the default apps, which are the only ones I'm reviewing, are also by Opera.

The analogy that jumps out: Unite would be the iPhone OS. The Apps would be just like the Apps for the iPhone. You need both (and indeed, the iPhone is evaluated as a whole, with non-Apple apps receiving their own reviews but Apple's apps are just part of the package).

That out of the way. Let's dive in!

Why are things the way they are?

When Dr. Eng speaks of the "The Internet’s unfulfilled promise" I can't help but applaud -- it is a worthy vision. We are so willingly surrender our rights when we press "I Accept" on click-thru-contracts. However, you cannot neglect what we get in exchange. In terms of usability I get access to services that help me do things. What sort of things? I can send pictures of my ugly mug to my friends, they can send songs to me that they love, I can update my good friend in Armenia that he got mail from the DMV recently (he can send back a forwarding address). All of these are forms of communication.

That is a fundamental of the Internet. As human beings we are constantly looking for efficient and effective means of communication (everywhere). We are naturally selecting the method which takes the least amount of work and is still effective. How many of us bother speaking when a wink will suffice?

The uncontested king of the Internet? Email.

That simple service changed the world forever (well... for as long as we get to rent the world). The question "Have you heard of E-Mail?" has been replaced with, "Do you have an Email address?" and in turn, "What's your email?"

Email is a very basic form of asynchronous communication that is relatively easy to use with a minimal amount of training. If you can send postal mail, you can master e-mail.

Postal mail is asynchronous communication. I drop a letter in any postal box and eventually it lands in a mailbox for the recipient to read at their leisure (or shred, depending on what they think I'm asking for).

The architecture of email is the same. I can send off a message and eventually my friend gets it. However, in his village in Armenia there is no Internet. He checks it when visiting a neighboring town every week. The beauty of email is that he connects to his mail server and it takes care of getting the message to me. Likewise, my email response will be held by his mail server until he can login and view it.

Now, you could run your own mail server but mail servers expect other mail servers to be "always on." If my friend and I ran our own mail servers we'd both have to be online at the same time in order to send and receive email from each other.

For most of us email works brilliantly.

Synchronous Communication

Unite is synchronous communication. You can only share when both "people" are online. It's more like a phone call to my friend than a letter. I consider phone calls way more fun than emails (usually) but they take more work to coordinate (not to mention, they can be pretty inefficient).

Unite takes more coordination, too. Every user who expects the level of functionality granted by email, by Facebook, My Space, Flickr, Dropbox or any number of the thousands of server-based-services that help us meet communicate, will find Unite lacking. Unite is competing against these solutions in the desires of common users.

But what's in a server?

Then there is the last-mile. My bandwidth could barely handle those 40 pictures. If I had conferenced called my friends (or even just sent out the link to everyone at the same time) what would the result have been? Very slow going for all of them as I reached my upload connection limit (512 kbps, I have found out).

These problems won't exist for everyone. Anyone who has a fast, dedicated Internet Connection with a computer they always leave on has an ideal candidate for Opera Unite. In other words, they have a server.

The reason we call computers servers as opposed to just computers is mostly semantic. We mentally categorize them as more capable of content distribution on the Internet. They have faster processors, usually some redundancy, faster connections to the Internet and they are "always-on". Unite says, "Any computer can be a server!" -- Yes, this is true! ... Well, when it's on. We humans depend on "Yes, almost always," for the solutions we'll continue to use.

That's at the heart of usability. When I give someone an email address the expectation is they will, at their leisure, be able to email me. Likewise, when I wanted someone to view my pictures, I need a service that allows them to do so at their leisure.

For the users who try out Opera Unite with the standard apps the picture Opera paints is very unclear. Read the introductory statements to Opera Unite (http://unite.opera.com), as one of many examples "Opera Unite allows you to easily share your data: photos, music, notes and other files." But missing in bold at the beginning of that statement is all the "Ifs".

The Ifs

If they don't always have Opera on, it doesn't work. If they aren't always connected to the Internet, it doesn't work. If they have a very slow upload connection, it doesn't work.

A supporting story: The friend who introduced me to Opera Unite (encouraging me to try it) also introduced another chap who quickly ran into problems. This chap had been trying to figure out why his friends couldn't access his files. We began trying to troubleshoot when he realized, "Oh, no one told me Opera had to be open."

Another, "duh"? -- It would seem so but it depends on where you stand. Are you a computer engineer or a computer user? Usability must remember your target audience, the whole experience and that they won't be using your product in a controlled environment. Look at fire alarms, it's always shocked me how easily, in some instances, one can pull a fire alarm and disrupt thousands of people. Yet... the trade off is usability, any attempt to circumvent pranksters would result in a very bad side-effect: People panicked by fire might not be able to figure out how to set off the alarm (or worse, not bother trying).

Other services are so easy to use and so powerful because of that basic simplicity - it always works. Unite doesn't lay out this restriction to begin with. Unite isn't a solution for asynchronous communication -- it is only usable when you are connected with those you wish to collaborate with. Fair enough, but say so up front because it is a big "if".

It comes down to usability...

Unite's promise is, "giving you greater privacy and flexibility than other online services." But other services work more (thanks to being inherently asynchronous). Facebook is the living proof that users are willing to sacrifice a great deal of privacy for effectiveness. Not that Facebook's power doesn't disturb me but let's face it, Fb is now bigger than many nations on the planet. Part 2 showed me conclusively it was more work to try and synchronize schedules, leave my computer always on, get a faster upload speed. But that design issues are not addressed at all by Unite.

And to date, my research has revealed no way of solving that problem. Sadly for the future of Unite, other server-services don't even have to contend with that problem. Those "big servers" leverage an economy of scale that we don't have access to on an individual basis.

The general usability of Opera Unite is severely undermined by the dependence on synchronous communication and further by a general expectation among those on the Internet about how file-sharing works (which Opera doesn't correct in it's "sell"). For chat, it's nearly perfect -- as chat is synchronous communication that doesn't take much bandwidth.

There very well may be creative workarounds to these problems but Unite's first impression is going to count the most for many people. When I need a quick way to make a temporary share of fresh data, I can see Unite as the solution. Perhaps developers will come up a killer app that really shines in the niche that Opera is taking.

That's where I see Unite ultimately landing, a niche. There are those who spend their days with their computer on, the browser open and have very fast connections (Universities?). This is fine and good but I fear it won't be the revolution that I'd dream of, it will not unite us all.

Opera Unite Review - Part 2

,

Now onto the use cases. I actually wrote this first and will leave it in that original form for the sake of preserving my perspective.

Opera Unite is still in Beta but the architecture issues here are still fair game.


I've been playing with Opera Unite over the past couple of weeks, using it to share files and I must admit I'm disappointed. What is the problem Opera is trying to solve here? Let's use Opera's own words, "Opera Unite is a form of collaboration technology that allows you to share data such as files or photos with other users, directly from your computer." (What Is Unite)

File sharing. Excellent! I love file-sharing and often long for an easy fast method of sharing files... We'll get to an overall evaluation after a case-study.

Here is my need: I have 40 pictures I need to share. I need my friends to be able to view those pictures, inform me of their favorites (and why) and keep track of that information. I'm trying to come up with the overall favorite picture and two different genre "subfavorites" based on their collective opinions (with some people's opinion receiving more weight than others)

What do I have? I have the 40 pictures on my desktop. They are high resolution images of ~4MB each. I have a relatively slow DSL connection (1.5 down/740 kbps up) but I'm not running any other services on my connection so that isn't too bad for my friends.

I have a few options.
  • I could email the pics, I've tried this and while it is very convenient for my friends, it's nearly impossible when dealing with this many pictures of this size (I could scale them down, but again... more work).
  • I could upload to a web server. Currently, I have access to only a few and they aren't my "personal web space" -- so I'm not keen on putting my personal stuff up there. Hunting down a web host doesn't hold much appeal either (nor the cost of it).
  • I could use a service like Dropbox, it allows me to upload to their server (up to 2GB) free and I just have to "drop" my folders into my Public Folder (it is integrated with my OS). Then my friends could goto that site and download and view at their leisure and not dependent on my setup. Not too bad but...
  • Opera Unite. Being an Opera fan, I decided to give Opera Unite a go.


Case 1: Setup & Go

So, I activate Opera Unite, track down my folder and set it to share. It takes me a while to find the password box (my window was shrunk and I didn't think to scroll down, I thought the password option would be in Opera Unite settings, not on the page itself). Which gets to my first issue. Settings are not centralized. There are the right-click Opera Unite options (this is where you set the folder you want to use) and then there are the "page view" options (this is where you set the password). I had to mentally bridge the gap ("Oh, there are two entirely different places where you can set vital preferences on Opera Unite.")

I got it setup quickly as I was racing out the door to meet some friends I wanted to show the pics to. I arrive at their house and was excited I was going to be able to just pull the pictures up!

I goto the webpage and I discover two major problems. First, I had set up the Opera Unite service for "File Sharing" instead of "Photo Gallery" -- Whoops! -- Now instead of thumbnails, I had a list of 40 files (and no way to change it remotely). Worse, I discovered how painfully slow my upload connection was. About 3 minutes, per file. That's a problem, because my 3 friends are standing over the computer and we want to browse through and compare the best.

Okay, I'll just download them all and then look at them... Surely there is a... Nope, no way to download all the files. That's the second major issue. I resign myself to clicking each file (but not too many, because it errors-out) every few minutes. Eventually the files get downloaded and I show my friends.

Case 2: Different Times

Next, similar situation, now long distance. I want my friend on the East Coast to look at the photos and give me her opinion. I give her the URL and password information and say, "check it out."

She sent me an email not long after trying saying she couldn't get it to work. It so happens I have my laptop off at night (dang fan is way too loud). Now this is a "classic duh". Obviously, if you're computer is off, it's not going to work but this speaks to a major problem with this model. Asynchronous file-sharing is not possible. With stationary desktops, this isn't too much of an issue, with laptops it makes the solution completely untenable.

Case 3: It worked great! and can I....

Next up, a different friend in the Central Time Zone. I was going to go through the photos with her on the phone. By now, I've setup the Photo Sharing option so that thumbnails appear and it's much easier to navigate the photos. It works beautifully (it's a little slow but not unbearable). We spend half-an-hour on the photos and I get her favorites, we go back and forth, comparing, narrowing down. Excellent.

All of a sudden, I realize there are 20 other pictures I want her to look at. Now -- I'll be honest, my immediate mental solution may not have been the most efficient or effective but it was the first one that came to mind. I didn't want to change my file structure (dragging the 20 pictures from their folder into my "sharing" folder). So I decided to add another "Photo Sharing" Service (they are now called Apps). Now this process is unclear to me and I'm not sure I found the best method but I looked around for a way to just share another folder and couldn't find one.

I went to the "Add Opera Unite" service (after trying to add a "folder" and not having it display any "folder" selection options) and just added another Photo Sharing service, setting it to my other folder. Then I had my friend login to that service and we continued on. It is still very unclear to me how I should have done it "right".

Case 4: Got it working smoothly. It'd also be nice...

Finally, my friend in Mountain Time Zone (LOL, I just now realize I am traveling across the US). I was at a coffee shop far from home and plugged in my laptop to go through the pictures. It worked well but I realized how badly I wanted her to be able to provide recordable feedback on the photos so I didn't have to keep a little notepad of that information. Not possible. (even if I just typed in her feedback on each pic)

Case 5: Show me the pictures!

There is one intervening incident that took place that is worth mentioning. I was at a friend's house and I wanted to share a bunch of pictures I had taken earlier in the year. Again, large in size, about 40 of them. Alright, go Opera Unite!? I activated the Opera Unite service and pulled up the photos on my friends computer. My plan was to simply save the files that she liked on her computer.

However, it was running very slowly. I didn't understand it, we were on the same LAN together, it should have been operating much faster. At first I assumed that it was because all data was sent through Opera's proxy server in Norway (while this provide privacy and some additional security the cost is speed). Though our two computers were literally sitting next to eachother, it seemed each packet was having to go to Norway (via my slow upload) and back (via the somewhat speedier download).

I'm not sure what the issue was that caused the major slow down. Perhaps the proxy did activate and my theory is correct but I saw no way of *knowing* that. There is no indication, no flashing light, no status report.

Even in the photo gallery, saving files was not obvious at all. Right-clicking on the displayed image showed "void.gif" -- I'm sure it had something to do with how the navigation worked but if I'm sharing photos with my friends, 9 times out of 10, I want to let them keep them.

I quickly gave up, activated Windows File Sharing (gah) and luckily it worked. I dropped the whole folder in her computer and 6 minutes later it was done (She can prune unwanted pics later).

What went splat?

From my usability perspective:

First, the UI is bit unclear. I quickly got the idea of "services" (and I applaud the change to "Applications") and how to turn them on. However, the division of settings means I have to search different locations in order to make modifications (there are even further settings in opera:config but you'd only know that if you Googled Opera Unite and read through the FAQ). Further, I couldn't figure out how best to share multiple folders in different locations on my HD.

Second, no easy way to handle big bundles of files. Remember, the problem Opera is saying it wants to solve is "File Sharing". Sometimes people want to share a lot of files but there appears to be no elegant way of handling that.

Third, I couldn't fix my "wrong service" remotely. I wanted Photo Gallery but selected File Sharing (you can see why I made a mistake, File Sharing is broadly what I wanted to do -- I just wanted to share them in a different format). This is probably good in a lot of ways but all I could think about as I had my three friends looking over my shoulder was "OMG! How do I fix this!?" From a security perspective I see no advantage in not allowing the remote change of how the files you are sharing are formatting.

Fourth, asynchronous sharing. This is a tough one, because the idea is to make your computer a Server. It works as designed. However, that design means portable device users must make their devices available to share. This presents a lot of logistical problems that the user has to solve to use OU when there are *far* simpler asynchronous solutions available (Hi DropBox).

Fifth, as it is explained on OU it attempts direct connection and when it can't, it goes through a proxy. I'm not sure what happened when I attempted to share files on the LAN but the connection was excruciatingly slow and there was no way to troubleshoot it. (Even just seeing that the proxy is not active would have allowed me to figure out what on my end was slowing the LAN link).

Sixth, the last-mile is still an issue for many users. OU might work great when you're on your University LAN or if you happen to have a high-speed upload connection but for the rest of us, are our friends willing to wait? (idea: Opera Testers signs on to OU via a local dial-up connection and shares with their friends, rate experience)

Some of these are easy problems, some though, are symptoms of deeper issues.

The thing to remember about usability is this: It must work for nearly everyone. It doesn't have to work the same way, but if 30% of the users can't make something work, it's a failure. I'd argue, that for wide adoption, you probably need 90%. Think about innovations in the world, even if they are complex, if people find the payoff worth the investment of time then it's "worth it."

Look at cars and their drivers. There are legally stipulated levels of operational ability to use a vehicle but the payoff is you can go most anywhere within certain geographical constraints. (My "It'd be nice" are the cup-holders of the world)

Likewise, email isn't exactly intuitive but the payoff for learning how to use it is cheap, immediate communication with the whole world -- most importantly -- it works well once you gain that basic understanding. Like driving, once you get it you are set forever!

Next up... back to the high level evaluation.

Opera Unite Review - Part 1

Greetings Usability Fans, Amateurs and Professionals!

I've decided, unwisely for myself*, to spend some time reviewing the new Opera Unite feature being tested and with plans for general release soon. I enjoy Opera and for the amount of work being put into this new feature it was time to give it a go and being the Usability Amateur that I am, I wanted to share my experience.

I'm going to work backwards. The first thing I did with Opera Unite was use it. I did not read about it. I heard from a friend that he had used it and it "basically turned your computer into a server." I think there is great value in this approach because it's likely the way most users will be introduced, via word of mouth.

However, for this series I'd like to start with where I ended, that is reading much more extensively about it. That is reading Opera's stated vision on Opera Unite (OU).

The ultimate description I found is found here: http://labs.opera.com/news/2009/06/16/ written by Dr. Lawrence Eng. To summarize, the Internet is becoming controlled by massive corporations with massive resources to dictate our usage of the World Wide Web (for better or for worse). This is not the original vision of the Internet, which relied on distributed computing to "do work".

Opera Unite aims to get back to the roots of the Internet by providing a platform that Developers may use to allow our computers to act as more than "stupid terminals". Casting off the restrictions imposed upon us by Service Providers who have billions of dollars at their disposal. In other words, the idea here is to give the power back to the individuals and their close "real" friends (as opposed to, say the 800 friends you have on Facebook that you may or may not talk to once every 10 years).

Much in the same way the iPhone revolutionized the mobile phone market as a legitimate and powerful platform, Opera Unite's goal is not to be just a web browser software company but a way for developers to step up and harness the computing power of your computer in meaningful ways.

Now, there is a lot of word semantics involved here but essentially Opera Unite allows you to run programs (Opera calls them "Services" "Applications") within the browser. Right now the programs offered are simple, "File sharing", Photo Galleries, a Chat Room, a posting forum. However, the plan is clear and the implications could be significant.

Besides the examples I mentioned, you could create your own "Web Server", uploading your own HTML. There's an example of a Service that allows you to be your own OpenID Provider. However, as far as I can tell, you could do things like run your own Mail-server or any other program offered on the Internet.

It is a grand plan, like the iPhone, to change the landscape of Internet computing. However, unlike the iPhone, there is a fatal flaw.

*Unwise because I should be job-hunting, this, though satisfying intellectually does not pay for the electricity bill.

Proclamation: I will never use a browser that doesn't support mouse gestures as long as I live

,

(...unless forced to by work)

Post updated 17 Sept

Seriously- I love the irony of people haggling over milliseconds of JavaScript rendering times for pages, but don't bat an eye at the amount of time spent waste pushing a mouse pointer around the screen to find the "back" button, the "X" on the tab,the "new tab" button, or moving your hand back and forth from the keyboard to the mouse. I would think the user's "Locus of Attention" would at least factor in to the equation to some extent.

To me, reading a page and being able to have my browser respond to my muscle memory (without thinking of the UI at all) to close a page, and go back and forth in history is worth... oh, I don't know... maybe 1.2 gigawatts of JavaScript rendering time (Great Scott!) The best UI is one that doesn't get in your way.

Opera still does this the best of all browsers I've used, Firefox is okay, but the gestures extensions just aren't as "fluid" as Opera. But, since I just pointed out that I don't like to nitpick over a millisecond here or there, I think it's perfectly fine for daily use.

It would be nice to see Opera (the only browser I know of that has Mouse Gestures baked in by default and part of the browser) to do a usability or eye tracking study to measure out "Locus of Attention" in manipulating the UI

Yes, Usability includes being able to use it.

, ,

I just read Network Woes? Hate the iPhone, Not AT&T by Mr. Newman. The basic premise: Don't hate AT&T because they're network can't handle iPhone data traffic. He points out that iPhone users use more and other networks don't have the problems because they don't have the iPhones.

Nevertheless, I contend it is still AT&T's fault.

Put on your time travel hats and let's head back to December 1996. America Online (AOL) is the leader of the online world by virtue of being the largest ISP. They've operated on a "hourly billing" system (yes, you used to have to pay-per-hour for Internet Access) and then they offered a new "unlimited hours" billing plan. WHAM! For months accessing AOL was like trying to reach the really popular girl in High School after her latest date. You begin to sing harmonies with the busy signal.

It was a classic case of a failure to appreciate the market and then worse, failing to communicate a failure. They removed the "scarcity" created by keeping the hourly timer. AOL avoided acknowledging the problem for a while and then avoided answering direct questions about how and when they were going to fix it.

They also reaped the rewards, membership and subscriptions skyrocketed... millions of people signed up and wanted to get online. Only to find it was America OnBusySignal. The decision was clearly "profit-oriented" and then the failure to take effective corrective action after they realized they were overloading the system. Yes, eventually they caught up with the demand but not after millions of people felt betrayed by the ISP. AOL could have avoided some of the upset by stating up front, "Due to overwhelming demand access to our systems may be limited until we are able to meet demand." -- yes, before someone signed the contract. Or... even, limit the subscriptions you accept to account for what you can actually provide.

Put another way, AOL oversold the movie theater, by a lot. They knew they didn't have enough seats but all they saw was dollar signs, so they sold tickets to everyone who walked up, knowing that they couldn't possibly get them all into the movie theater.

It should be considered a fundamental tenant of usability: You must be able to use it (in this case sign on).

Perhaps this is a flaw in capitalism. It's far better to take that risk, over-promise and not deliver then to be more conservative and lose customers (aka "leave money on the table").

When you walk up to a movie theater and you buy a ticket, you expect to see that movie. When you buy a plane ticket, you expect to have a seat on that plane*. When you signed up for AOL, you expect to actually be able to use AOL. And finally, when you sign up for an iPhone and two-year contract you expect to be able to use it.

It's AT&T's responsibility to provide service capacity or warn customers prior to signing the contract what the service REALLY will be.

Again, it's important that usability includes being able to use it. This means capacity issues must be taken into account when designing systems. This means monitoring your capability to provide consistent, reliable access to your customers.

However, history seems to dictate we'll continue to have this situation where developers fail to take into account "the big picture". It's a complicated issue, including system redundancy and sophisticated network management.

Similar examples. Gmail's recent Email Outage was exactly the same type of failure event as the Northeast Power Outage of 2003. The redundancy system worked correctly but designers had failed to properly take into account the failure of multiple nodes across the system that automatically redirected demand to others, causing them to fail in turn. (If anyone out there has a technical name for this problem, I'd be curious to learn about it. It seems like the combination of a Feedback Loop and Domino Effect. Domino Feedback Loop. :wink: )

In short, the capacity problem that AT&T is facing is not unheard of and could have been predicted (or at least dealt with differently now). So, who's fault is it? That'd be AT&T. The blame is where it belongs.

Yum
* Yes, I am aware airlines oversell their seats, I used this example specifically because it establishes a system for dealing with their oversell. You are financially compensated if you can't take that flight.
** I don't have an iPhone, I have a Gphone. -- T-mobile has it's own issues.