This site is rarely updated. benbrophy.com is more up-to-date. - Ben
New high in Stellar usage
I just got this from Craig.
On Tuesday, we hit 500 users by our metric. We had 2850 unique users logging in that day, 2605 wednesday, 2381 thursday.
Those are some big numbers for us. I was using Stellar at the same time as the 500 on Tuesday, and I didn't experience a performance hit, so the servers are handling the load nicely.
QA and Iterative Development
Carol Dippel is doing a presentation on improving the QA process for Sakai using CMM methods.
These methods assume formalized documents through out the product life-cycle. In other words it assumes the QA team will have details on what the application is supposed to do before before the test to see if it does it. Sounds obvious, but it often doesn't happen. Writing specs takes a lot of time, and it's not fun, I've been as guilty as anyone here.
There's a tendency to add in new undocumented features at the 11th hour, because it turns out to be possible. This is hell on QA. She'd rather have the newer stuff slated for a later release, say 30 days later.
She's not advocating a slow bulky 'big bang' approach with massive application releases planned a year in advance. It's that you can have small monthly releases with iterative improvements. This makes QA really workable, because the QA team can find bugs more quickly.
This sounds like an 'extreme programming' approach, nicely applied. Carol doesn't call it that b/c people use the words 'extreme programming' to justify a chaotic unplanned software development (note to self: check out this extreme programming book).
There's a lot more going on in this presentation - this was just one theme I thought was interesting.
Podcasting
I finally understand what podcasting is after reading comments on a blog entry.
Ruby on Rails
I am constantly seeing links to Ruby on Rails. I'm tempted to spend some time trying it out, but I don't have the time. Maybe I'll ask one of the mirror-Ben's in an alternate universe to spend some time on it.
Generating actors and Tasks
I was in a meeting today using a some "High-Tech Anthropology" techniques to generate ideas for an improved resources tool for Sakai. The the most important goal of the meeting was to gather our input fast.
Here's the process:
- Quickly write up a brief problem statement in small groups
- Brainstorm a list of 'Actors' who interact with the tool (e.g. Instructors, students, librarians, etc.)
- Brainstorm a list of actions the actors would do in using tool (e.g. import last semester's class, search repository for class materials)
- Categorize and refine those lists
- Create a matrix of the two, and in date how often the actors do the tasks (e.g. instructors export sites once a semester, students download files hourly)
- Vote on which actions we think would reap the highest benefit for our users
- Look at the results of steps 5+6 to pick a few actions to start with to create use case scenarios.
The meeting was really productive as we went through the 4 stages. I thought we generated a lot of information and ideas fast, and we started forming a rough shared vision of the task ahead.
We lost focus as the day wore on - we were getting rushed, and there was a bit of confusion over our process, especially step 7 when we spent a lot of time messing with an excel spreadsheet and slowly realizing we were treating qualitative information as if it was quantitative and it just wasn't working.
The day a real success, despite the seventh inning slump. We generated a list of clear requirements for the lead designers, and they also got significant buy-in from the rest of the team. I'd love to try the first 5 tasks the next time we have a client who is sure how to define their project.
Create your own bitmap font
BitFontMaker looks so cool. I can't wait until I have a free hour to really obsess over it.
Trackback Spam
My family blog was attacked by dozens of trackback spams today.
- Trackback
- Trackback is a cool technology for communication between blogs. example: You write something interesting about cheesecake on your blog. After reading your entry, I link to it on my blog, with the additional comment that I think chocolate cheesecake is the best. My blog sends your blog a 'trackback ping.' Then your blog has a link to my blog entry.
- Trackback Spam
- Trackback spam is when you write a blog about cheese cake, and I send you a bogus trackback ping about an online poker website, so that your site will show a link to that site and increase it's google rank.
I'll probably turn off track back on my site now - because no one has left me a trackback on my site since the early halcyon days when there was still a blogging 'community' (back in 2002). Now trackbacks are for scum leaving nasty links around.
It's like People's Park in Berkeley. I read about this amazing park and garden built by volunteers back in 1968. I walk by it on my way to campus now, and it's full of scary desperate people - you wouldn't feel safe walking through it at 9:00 am, let alone at night. Sometimes a free and open community attracts first the best and then the worst from humanity.
(this blog takes neither comments nor spam, which has it's plusses and minuses)
Javascript
This tutorial on unobtrusive javascript looks great. When I first saw those tacky 3D logos I thought I was wasting my time, but the examples are perfect, and the 'unobtrusive' philosophy sounds right on.
Back in 2000 or so I despaired of of javascript ever being useful to me. The browser interoperability issues were too great. When I use it in a project it's either copied whole sale from some site I've googled, or i've cajoled a programmer into doing it.
I know if I learned javascript I could use some very handy UI tricks. I loved playing with ActionScript in Flash, so I imagine I'd enjoy using the current standards based brand of Javascript. Someday I'm gong to go back to javascript school and learn it right this time.
Wireless Access in Philadelphia
The NYTimes has a good article on plans for broadband wireless access in Philadelphia. I wrote a letter a while back to the Governor of Pennsylvania urging him to veto a bill the PA legislature had passed banning the local governments from providing internet access. Obviously bill was written by lobbyists working for the telecommunications industy.
I'm all in favor of interent access becoming a commodity, available to everyone at low cost. It's interesting that the companies involved aren't interested in providing ubiquitous wireless access themselves, they just don't want the city doing it.
**Update**: looks like Indiana is doing the right thing too.Not responsible
We use basecamp to track our projects, including the Sakai Gradebook. Today I went in to the gradebook to-do list, filtered to see just my tasks, and checked off the tasks that remained. After I checked off the last one, this message appeared:
Ben Brophy isn't responsible for anything
How satisfying!
Gradebooks are complicated
We had a long meeting about what happens to unscored fields when an assignment is released. It effects what the student sees, how the final grade is calculated, how the average scroe is displayed, etc. I changed 6 wireframes in the gradebook specs after the conversation.
I always wondered why most course management systems either lacked a grade book or had one that was airly lame. Now i know that it's because they are really complicated. There are so many different styles of grading to account for. Even when you stick to one style, there are lot's of logical implications to changing the way any one number is calculated.
It's going to take two or three releases to get back to the full functionality we'd hoped for. Right now I'm glad to be implementing one fairly simple grading style, and using this first try to work through all of the weird kinks of logic.
Potential Docwalla User
Wow. Somebody just emailed me about Docwalla (Simple XML/XSLT/PHP Micro-Content Management). They apparently have every intention of using it. That would make me so happy.
I'm still working on Docwalla... verrrrry slowly. Every to days or so I spend an hour working on it in the evening. I am slowly making the wishlist truly cool (it's already kinda cool). I got a lot of milage out of my wishlist this Christmas, but i got some feed back from both list managers (Lisa) and end users (our families) about improvements we should make.
Automatic Upgrades
A question came up in the Stellar team about whether we should upgrade old project sites (those sites that aren't ties to a semester) to the latest version of Stellar. We could do the upgrade or we could email them all asking if they wanted the upgrade.
This is a case where I think less choice better. Why ask? Just upgrade them. By asking them to decide and email us we're giving busy people another job to do (not to mention the extra work for the person who has to selectively upgrade sites as the requests come in). I do think it's worth emailing them that the improvements were made - for PR value if nothing else.
I've been using flickr.com for photos. The other day they upgraded their UI, changing the way slideshows work, etc. I haven't heard any one say "Damn, why'd they improve my site with out asking me first?"
Attention RSS Feed Readers
I am moving to a new RSS format. The default that came with Blosxom was the old RSS 0.91 (yuck!). I have upgraded to RSS 2.0. Here's the feed you should use:
http://web.mit.edu/benbr/notes/2005/index.xml
Please switch over. The old RSS feed will no longer be updated. Sorry for all the shifting around, I'm still settling into my new home here. There is a 90% chance this is the last time I'll change the URl until next year.
Also, please take a moment to click the link and admire the attractive CSS styling. Much nicer looking than most RSS feeds you've seen, isn't? If you use a PC, please shoot me an email and let me know how it looks.
Access and User Groups in Stellar
Access issues are the thing we get the most questions about in Stellar. I'm thinking about making it more obvious in the user interface somehow.
I wonder if there is away to bring access up more easily, but I think figuring out how to grant access is fairly easy (if improved access is a priority for the next Stellar update, we can check with usability testing). It may be what is needed is away for people to request access.
Most of the reports we get are from people who say they are being denied access unfairly. There are students who aren't on the registrar's list. OCW, SMA or Library staff who are helping of a site. Instructors who want access to an earlier version of the class they are teaching. They have assumed that the developers control access, when really they ought contact the instructors or TAs. So perhaps we need a form they can fill out to request access right in the place where they get the bad news - the permission denied page.
Instructors could then be emailed, or be greeted with the request when they next visited the site. This might make life easier for everyone. People looking for sites won't waste time contacting support. Site owners don't need to find the right form - it's presented to them when they need it. And (this part really delights me) the support and development teams will explain access less often.
Update: I got some feedback from Keith at OCW. Based on his comments, and embroidering them with some new ideas of my own, I'll add a couple details to my plan.
The permissions denied page should only show the 'request acces' form to people who are part of the MIT community so instructors don't get bombarded with requests from the rest of the world.
The email instructors get will contain 3 links:
- Give [requester's name] guest access
- Give [requester's name] higher access
- Deny [requester's name] access
The first and last link would just check the certificate, do the job, and confirm. The middle link would give more options. That way the instructor can process the email in just a few seconds.
Don't Make Me Think
I am bummed that I'll miss Steve Krug talk at MIT this week (not to mention Joanna's birthday lunch - big week in Cambridge). Don't Make Me Think was one of the books that really helped me when I was a newbie, and I'm pretty curios to hear what he's got to say now.
Steve Krug, Author of "Don't Make Me Think" Steve Krug was a little-known but highly respected usability consultant until he wrote Don't Make Me Think. This book, written with wit and much common sense, is a product of more than a decade of user advocacy with companies like Apple, Netscape, AOL, BarnesandNoble.com and others. Wednesday February 16, 01:30-03:00pm, 3-133
Better metadata for Stellar materials
Classes from Stellar can be imported into OCW. OCW uses a different nav bar. That caused a problem: On the imports the OCW workers had to figure out how to sort a big pile of Stellar 'materials' into their categories of kinds of documents.
To help solve that problem we made the 'kind of document' metadata options for each document added to Stellar match OCW's categories. But instructors weren't bothering to change the default - "readings" - so the problem wasn't solved.
We then made it possible to use the OCW nav bar in Stellar. This presents a usability problem though - students have too many places to look to see if new materials are added. Not a problem in OCW, because the sites are static. But a Stellar user needs to visit the site regularly, and they shouldn't have to comb through many pages looking for the new stuff (the new Stellar RSS feeds will help solve problem). Instructors again chose to stick to the Stellar default.
Last semester we started requiring instructors to pick a 'kind of document' when adding something to materials. There is no default, so they have to choose. I felt a little bad about this, because it slows them down. But guess what? It works!
I just picked 5 Stellar classes at random (admittedly not a big sample when there are 383 class websites for Spring 2005 but I have other work to do).
- 3.044 - Materials Processing
- BE.010 - Intro to Bioengineering
- 9.02 - Brain Laboratory
- 12.007 - Geobiology
- 11.432/15.427 - Real Estate Capital Markets
Each of these classes marked their materials quite nicely. They have a nice diversity of 'kinds of documents.' 9.02 had mostly readings, but on inspection they really were readings. Check this out: every single one indicated which document was their syllabus. That's a big win right there.
3.044 was the only one to use the OCW nav bar, but interestingly they also use the all-inclusive materials page. I'm not sure that's a great solution for usability (that nav bar is way too long) but perhaps they feel they are getting the best of both worlds. I don't think they classified their documents any better than the others. They all did a good job.
Just little user-generated-metadata success story.
iPhoto 5
I installed the new iLife '05, mainly so I could check out the improvements in iPhoto. The biggest immediate for me is iPhoto now imports the sort movies I can take with my digital camera. I used to have to use iPhoto for the photos and Image Capture for the movies.
There are also a lot of nice UI tweaks, for example when you go into a close view of an image, there is now a row of thumbnails for the other images in the roll at the top of the window. Clicking 'Done' or clicking another roll brings you back to the catalog view. That's swell.
Another nice touch is the 'enhance' dialog, which is much sexier version of the 'adjust levels' (and brightness, and contrast) that you find in PhotoShop.
There's also some bad news - Export wasn't not working, perhaps becasue I've got a Flickr plugin. After I removed the plugin and restarted iPhoto all was well. I need that plugin though! I've got my fingers crossed that I can find a new Plugin on the Flicker website tomorrow.
Update: I downloaded a fresh version of the FlickrExport for iPhoto and it works swell.
HumaneText
At some point I need to try out HumaneText to speed up my XHTML writing. Might be especially handy for this blog.
[See the discussion on 43 folders]()
Update 1: I'm trying it now. Can't get it to work easily. HumaneText (the service, I haven't tried the 'unix filter' mentioned in the comments above) doesn't seem to work at all BBEdit, and I need to try 2-3 times to make work in Text Edit. Doesn't seem like a miraculous time saver so far. I was really excited for it though - it's a great idea.
Update 2: OK, I got it to work well in BBEdit instructions on Daring Fireball. I'm using 'command+shift+>' as the command. I'm not going to set it up as a Blosxom plugin, because I'm so used to writing in XHTML it still feels slower to write in Markdown. I'll try it for a while and see what I think.
In my next actions list: write a markdown cheatsheet in my moleskine to use until it's a habit.
Scheduling updates with crontab
I just set up my laptop to do these blog updates for me automatically every 30 minutes. Here's how I did it.
- In the terminal type: crontab -e
- This opens 'vi' a commandline text editor. ESC i to edit
- I added this as the first line: */3 * * * * root /usr/bin/blosxom.cgi -password='mypassword' -quiet=1
- escape key, then ':wq' to save and leave vi
If you are reading this, it worked!
Google Template
I just added google search to this site, and customized the results page. It was quite easy. I was surprised to see that what method for customizing is a XSL file. You really get a lot of control. I'm delighted.
I've MIT's old search capabilities were nice but it was near impossible to customize the results pages. The new goolge search is going to e great for our clients at AMPS. Now that I now how to do it, I'll have to customize Stellar results pages.
Check out the handy IST info site on Google Search at MIT.
Simplicity
republished from my old blog
I'm moving to a new blog. My apologies to anyone reading, I hope you'll add my new RSS feed to your reader. If you like to read using an old fashioned 'web browser' you can visit the website, now called Inline Comments (it's a geek site, what do you want?).
My new blog is powered by Blosxom. I publish static files from my laptop over to my AFS locker on MIT. This means there are no comments or trackback. I have chosen to strip out a lot of other stuff as well, for example the home page displays only the most recent post. In fact, as I mention in the sidebar, the website is really just a life support system for the RSS feed.
Here's why I did it: Simplicity.
I'd like to continue publishing my notes, for my own record as much as anything, but since I am doing this from work, I can't devote a lot of time to it. With the setup, I write plain text notes and save them in a folder on my laptop. A chron job runs every 30 minutes and adds new posts to the blog in the background. If I write posts when I'm offline, they get published the next time I have internet access.
This is a fast efficient way to publish. As much as I love love love getting comments, they are another channel of communication to keep track of, and I've got a few to many of those. Just email me (note to self - add a cgi-email form to the site). Or blog it yourself, if I know you I'll see it in your RSS feed.
Latte to Go
There's a place in Berkeley where I buy a latte on my way into the office each morning. Once thing I really like about is you just ask for a latte. There are no sizes - not even small and large let alone 'venti' and 'grande.' The only choice involved is 'for hear' or 'to go' (or, since we're in California, 'para var').
Sometimes it's really satisfying to have fewer choices.
Styling RSS
Soon, I will add this tag to my RSS feeds right after the XML-processing instruction:
<?xml-stylesheet type="text/css" href="rss.css"?>
And create a stylesheet that makes the RSS more human readable. Ideally with a link about how to use it.
Earlham College and Sakai
I got my Alumni magazine yesterday and saw that Earlham is now using Moodle as their Course Management System. I looked at Course Management @ Earlham website and was startled to read they are also piloting CHEF/Sakai.
Their site has student, faculty and administrator evaluations of CHEF as well as a bunch of interesting reports.
Tom Kirk from Earlham listed several concerns with Sakai.
- Large universities are driving Sakai development, small schools like Earlham could get lost.
I don't think they should worry there too much. MIT and Stanford have a small school feel to them - mostly small classes with lots of faculty/student interaction. Having gone to Earlham and worked for years at MIT, I don't think the difference is so profound that the schools wouldn't use the same software. - Sakai is written Java, which they don't have in-house knowledge of.
A very valid concern. Not only is it in Java, it's in JSF, an obscure Java specialty. Even Java programmers have a big learning curve approaching Sakai development. - Weak internationalization in Sakai
It's hard to teach Arabic with Sakai right now. I know that the Sakai developers are working hard on this, and it's a strong requirement for the core schools as well - I'm sure it will be resolved soon.
If I was at Earlham I would favor Moodle as well. At least for now.
Indiana is hitting Sakai with 90,000 users this Fall, I don't think Moodle could handle something like that. Earlham doesn't have to worry about scaling too much because they aren't that big.
Moodle's got similar featuresto Sakai, more actually. Neither of them has a truly inspirational UI.
If I were running Academic Computing at a small college like Earlham I'd make the same decision - use the skills available in house to work with a PHP environment they can easily host locally. But I hope they keep an eye on Sakai, it is bursting with potential.