MIX09 - Day 2


Bill Buxton introduced the keynote today, which is about the release of Internet Explorer 8. The intro video, once again, was awesome. I think every web meme in existence showed up in this thing. (Not as good as ScottGu wrestling a bear yesterday, but pretty funny nonetheless.)

The first speaker is Dean Hachamovitch, GM of IE. He has some interesting points. We need a browser that “just works” for people who want to browse. Something that’s secure and stable. We also need a browser that works well for developers (uh… Firefox?). That’s what, apparently, IE8 is supposed to be. Available today, you can go download the final release version of Internet Explorer 8.

Some interesting statistics presented and the way they dealt with them in making IE8: 80% of user navigations are the user going back to a page they already were at. 70% of people have more than one search provider installed. To address that, the search box will return results as you type that come from your history and make that easier to get to. They also added easy buttons at the bottom of the search results box to toggle search providers on and off.

Oh, surprise: when a browser crashes, users don’t care why it crashed, they just don’t want to be interrupted. Not sure what genius figured that one out. The historic problem is you might be dong a bunch of stuff and if the browser crashes, you lose everything. To answer that, they did the thing Chrome did where each tab runs in its own process so if one crashes, the rest don’t. That took long enough.

Some of the performance statistics they’re showing are nice. Comparable to Firefox 3, nice and fast. Faster than Chrome. I’ll have to see if that plays out in more day-to-day scenarios.

Some of the little security stuff they did is nice. The top level domain is highlighted in the address bar so it’s easier to see. Say you went to “http://www.paypal.badguy.com/foo/bar/baz” - it’s not obvious that you’re not on Paypal… but if you highlight the top level domain, it is: http://www.paypal.**badguy.com**/foo/bar/baz. Oh, and built-in clickjacking prevention, that’s cool.

The standards compliance stuff is compelling… but the side effect of showing that IE8 is really standards compliant is that it shows the other browsers might not be quite as compliant, so you’re still going to be dealing with cross-browser formatting problems. It’ll be more compelling when all of the browsers get as behind standards compliance as this.

Web slices look like an interesting developer technology. The’re these little HTML snippets that run in a tiny gadget-style window in IE8 so the user doesn’t have to open a whole tab and log in. I can see some interesting potential use cases in some of our projects - let you get your account balances, for example, without having to go to your banking site. Examples they showed on this is the ability to check your Yahoo! mail or look at traffic reports in little web slice windows. Sounds pretty easy to implement, too - just add a few tags around exsting comments.

Accelerators also look pretty interesting. Context-sensitive functionality like the ability to highlight some text and send it in Gmail, or select an address on a page and get a map. That content, like the slices, shows up in a little gadget-style window. I wonder if it would be interesting to people to be able to, say, highlight a biller’s name and have an accelerator to start a payment to that biller.

He’s making a big point about the fact that “they’re going to listen to the users” in the future. Interesting. I mean, we all know they didn’t listen to us before, but dwelling on it shows they really heard that this was an issue. Let’s hope it sticks.

Next speaker is Deborah Adler, a designer who revolutionized the way pharmaceuticals get packaged and labeled. Not a techie by any means - not even someone who interacts with the tech world. She started out by trying to solve a problem for her grandmother and ended up solving a problem for the world. The problem was that her grandmother mistook her grandfather’s Alzheimer’s medication and took the wrong one. Same medicine, similar names… but different doses. Problems.

Other problems she saw were things like people chewing pills that shouldn’t be chewed because the warning about that got hidden or obscured among a lot of other text on the bottle. Apparently like 60% of Americans make mistakes in taking their medication because of difficulty in reading or understanding the instructions. Showing us the issues made it very clear - poor coloration, far too much text that is diffcult to read in tiny print, extra pages of difficult text to accompany the prescription bottle… I’ve seen it myself. It’s not clear. Tiny, poorly printed labels that make sense to the pharmacy but not to the end user.

Her solution - a revised label - is really good. It still has all of the information on it, but formatted in a much clearer manner where the information you need immediately (what the medicine is, how to take it) is prominent and the less important things (the phone number for the pharmacy) are less prominent. Labels get color-coded on a per-person and per-medication basis so my prescription for something will have a different color label than your prescription for the same thing - so I won’t accidentally take your meds. The bottle is reshaped to be flat on the back and slightly round on the front so you don’t have to rotate the bottle 360 degrees to read the information. Warnings go in bold, clear print on the back of the bottle. And a huge improvement - the label will actually get a red X that shows up on the front when the drug has expired so you know not to take it. Automatically. (Like time-release ink.) Standardized warning icons that are clear and easy to understand.

She tried to get it pushed through at the federal level but, while the FDA liked the idea, each state has its own pharmacy board so they couldn’t do it. In the end, Target took it and they’re using the idea now. It’s now called the “ClearRx” system.

This is really cool - it’s designed specifically for good human interaction. Granted, there were challenges in getting it out there (there are 23 different variations in the label to accommodate the different states’ regulatory requirements) but it’s a huge improvement from the crappy orange hard-to-read bottles.

Gonna have to ask Jenn if she’s seen this at the VA. The Surgeon General really likes it.

Makes me wonder what major changes we can make online to help people this much. Working in online banking, I’m sure there’s a lot of improvement we could make to clarify what people are looking at and make online banking easier and more compelling.

Wireframes That Work

Presented by a representative from a company called Cynergy that does contract RIA design, primarily in Adobe Flex. They list Bank of America as a customer, which is interesting to me.

Interesting point one - good design does not necessarily equate with good user experience. The example here was a house in Germany that won Time Magazine design of the year. It looks great… but the people living there aren’t having such a great time. Great design, great look, not great UX.

So here’s a new xDD acronym for you: Purpose-Driven Design. This seems to be the idea that you need to design your experience with the end purpose of the app in mind. Tailoring the experience to the user, the user’s needs, and the overall aim of the application.

Interesting idea that came up (that I happen to agree with) - don’t wait for the users to come back and complain about the experience before you start fixing the problem. Anticipate the issues and fix them up front. How often have you been on a project where you clap some UI on something that you know isn’t awesome but that’s what the stakeholders asked for… only to hear that it’s not the greatest and it needs to be redone?

Everyone comes into the deisgn process with some baggage - tunnel vision (thinking you’re limited by technology or “this is how we’ve always done it), changing minds (or not making any decision)… In a purposeful design scenario you have to step back from that and look at the problem. Watch the customer do their work. Look at the pain points. Look at the problem you’re trying to solve. Solve it without that baggage.

A tip presented: Turn off your computers when doing high-level design. Use a whiteboard. Use a pencil and paper. Computers are great productivity tools, but how many times do you check your email, get interrupted by IM, get sidetracked… and it’s true. I think about how I work and I totally get all of that information coming in all the time. (The computer will obviously have its place in the process, but try doing some of the brainstorming without it.)

And a note on process: Don’t be so rigid in process that it hurts the development effort or the flow of ideas. Hmmm. That’s definitely something I’ll have to take back to work with me next week.

From the presentation:

  • “It hasn’t been hard to make things look interesting or cool. Usefulness and joy can be elusive.”
  • “Design like an architect, refine like a sculptor.”
  • “Don’t be a usability nazi.” (This has to do with the idea of getting too caught up in process and letter-of-the-law usability guidelines like the Jakob Nielsen things like minimizing number of clicks and such. Solving the problem in the best way might break some of those guidelines but will actually provide a better experience.)
  • “In software, the desired goal is often a disruptive solution in the marketplace. Know that this may require a disruptive process.” This is definitely one I want to take back to work with me.

For my projects, I know I have opinions about how we should be doing things. I’m going to have to stop and think now - am I looking at it with my baggage-goggles on? Or am I really solving the problem? I know our UX folks are doing a great job at researching peoples’ needs, and I’ve seen the personality profiles and such that they’ve come up with… but one of the questions I have now that I didn’t think of before - have we talked to people who don’t do online banking and figured out why? Are we solving the people for only existing users or are we solving it for everyone? How do we solve the problem in such a way that we can increase our user base instead of just retaining the existing folks?

Lunchtime - Microsoft Surface

Got to play a bit with a Microsoft Surface during lunch. It’s sort of hard to really understand the coolness of the tactile experience without actually doing it. The videos and demos you see are neat, but when you actually use it, it makes a lot more sense.

One of the apps they had was a CD player where you set the disc on the table and it [somehow] looks at the case, figures out what the CD is, and starts playing the music from it. And, of course, you’ve seen the demos where the person sets their phone down and starts working with the pictures on it.

What if you could set your wallet on the table and see your account information? See your balances and such for your various accounts and credit cards? Want to pay your credit card bill? Drag a payment from your checking account over to your credit card account. Work with your electronic balances as easily as you work with cash, adding an easy to understand, tactile experience to your online banking. Might be interesting. Now if I can just convince work that I need to get a Surface… you know, for development purposes.

Securing Web Applications

I admittedly got here a few minutes late because I couldn’t find the room, but coming in… it looks like a better title for this would be “How We Improved Security in IE8.” Not quite what I expected. We’ll see.

Oh, yeah, uh… looking at the description - “Learn how to take advantage of browser security improvements to help protect your web applications and visitors.” Might have to go see what other presentations are out there. Recent projects have taught me that the security department won’t let us trust security to the browser - we have to control it all entirely at the server level. So…

Choosing Between ASP.NET Web Forms and MVC

This session is to help you determine what’s better for you - standard ASP.NET web forms or the new ASP.NET MVC framework. The demo shown here is two applications that have identical user interfaces, do exactly the same thing, but one’s web forms and the other’s MVC. Comparing apples to apples, so to speak.

Interesting bit when describing the way the demos were put together - a guy asked why there weren’t any themes used (.skin files, etc.) for the demo and all the styling was done in CSS. The answer - no web controls in MVC, so it doesn’t make sense to use .skin files. Interesting because I’m curious why it wouldn’t work if you were using ASPX as the view engine. Thinking what they meant wasn’t “you can’t use them” so much as “we chose not to.”

The presenter (Rachel Appel) seems to be dwelling on the URL format that MVC routing gives you. She brings up the querystring vs. nice routed URLs… but you can use routing with web forms. I’ve done it. Not sure the URL format is a selling point one way or the other. (Actually, later she mentions that routing will work with both, though she did pretty well omit it and sell hard when talking about MVC.)

She also seems to be talking about using web forms but NOT using the MVP pattern to separate the code out of the codebehind and into a separately testable class. I think that’s missing here. She brings up a lot about separation of concerns, but you can get some pretty good SoC with MVP.

I think the best part here and the most obvious thing that never gets said: With MVC you get full control over everything… but there’s a corresponding increase in effort to get results out of the box. You don’t get anything for free. Sort of the Spider-Man “with great power comes great responsibility.” Kudos to Appel for saying it. It’s true, and no one ever really mentions that.

Another thing she said that never gets said: when showing a <% foreach %> loop building a table, she mentioned how this is reminiscent of classic ASP. Absolutely. What she doesn’t mention is that the next logical step of creating lots of pages with tables is to create a block of logic that you can call and pass data into so you don’‘t have to write the <% foreach %> on every page with every table. Isn’t that… server controls?

Really this solidifies my thoughts that the best way to go is a sort of middle ground: web forms using MVP, taking advantage of the routing (which shipped separately from MVC, by the way), and having all of that third-party control support and the richness of web forms while also getting your separation of concerns goodness.

Granted, I very well could be convinced otherwise when MVC 2.0 ships, whenever that is. I was talking to Eilon Lipton on the MVC team last night about some of my concerns that never seem to be shown in the MVC demos. Complex input validation and localization. Can it be done? Sure, but it’s not really a great story. Again, with all that control, you get a lot more manual wireup and, in some cases, no help at all. Apparently some of these more complex scenarios are on the list of things to address. Looking forward to seeing that.

File -> New Company: NerdDinner.com

In this one, Hanselman is showing how to easily create a reasonably rich application, his example being a dinner scheduling application. Technologies used include LINQ to SQL and MVC. The data is getting abstracted away with the repository pattern. A very good demo of how you can really rapidly get something going here. Also a good overview of how MVC comes together. Probably a little more useful for the folks who haven’t messed with MVC, but good to see it all come together.

You know how you say a word so many times you forget what it means and it sounds like gibberish? The word “dinner” has been worn out for me now. Dinner dinner dinner dinner dinner. Yup. Meaningless.

New favorite site: sadtrombone.com. (Yes, you can find anything on the web.)

ASP.NET MVC - America’s Next Top Model View Controller Framework

This is an introduction to MVC given by Phil Haack. File -> New Project demo including a walkthrough of the project structure. How controllers get set up, that sort of thing.

I think this should probably have been given on day one to give the people a foundation on which to build over the course of the next two days.

Connecting Applications Across Networks with Microsoft .NET Services

This is an intro to the Microsoft .NET Service Bus, which looks interesting, particularly since we’re doing a lot of WCF in one of my current projects. Clemens Vasters is the presenter on this one.

Lots of interesting features here. For example, they’re working on a feature where you’ll still be able to connect to your service endpoint even if the port is blocked by the firewall. Sounds sort of like the way Google Talk will use port 80 instead of the standard Jabber port 5222 if it’s blocked. No real details but, still, on the horizon.

Another interesting thing - if you have a client talking to a service and the service bus detects that, say, they’re in the same subnet, the bus will detect that and upgrade the connection to get the client talking directly to the service. There’s an event you can listen to that will tell you when that happens. (I’m pretty sure I’m understanding that right, but I admittedly came in a little late.) You can also set connections to be reliable so if a connection breaks it’ll automatically be re-established.

They have a queuing behavior where you can send messages into a queue and the service will pull messages off the queue and respond to them. This is set up through a policy in the service registry. He made a big deal to say this isn’t, say, MSMQ queuing, but I’m not really sure how specifically it differs. The behavior seems to be the same, but with some REST sort of semantics based on HTTP verbs (like “GET” on the queue will read a message on the queue bot leave it and not dequeue it).

Something else interesting - if you want to see what’s subscribed to a certain message set, you can do a GET on a router subscriptions feed and get an ATOM document back with the list of all subscriptions. Do a POST to create a new subscription, DELETE to unsubscribe… all RESTful semantics around that subscription endpoint.

Good demo just sort of solidified it for me, though. Sort of like a chat app. Two Silverlight applications subscribing to a service on the bus listen for messages. Someone enters some text, submits it to the service. The service turns around and sends a message to the subscribers

  • the listening chat clients. Both chat clients get the text that was submitted. Basically Twitter. Got it. I see what’s going on now. (Oh, hey, the demo’s called “Text140!” I get it!) Was feeling a little out of sorts for a bit, not really knowing what I was looking at. Messages, at least in the demo, all take the form of ATOM entries.

OK. I get it. REST + ATOM + pub/sub + cloud = Microsoft.ServiceBus. Basically. Nice. Unfortunately, with the cloud portion, I don’t think we’ll be able to use it for the project I’m on (banks + cloud isn’t gonna happen) but I can see that it could be very useful in other scenarios. Twitter competitor? :) (Didn’t realize it was an Azure service until pretty late in the game. Again, probably from being late to the show here.)