'Tis the season...

Winter Break!

The last of grades have been posted and now it's time to clean up the mess from the last few months. It's been almost a blur since late summer, between make-up work, teaching four classes, grading for a fifth, and finishing my GPS project and paper. I'm quick to forget the difference between being awake and being productive as I stacked on week after week of sleep deprivation to try to catch up and stay on top of tasks.

But now, with a week of normal sleep and kids at home, things look a little different. I have one new class to prepare for (and possibly two), and more sections of things I've already taught. I went into the semester under-prepared, and had to make up for lost time at the expense of my own research.

Falling behind schedule just becomes the gift that keeps on giving.

Teaching can be all-consuming, and it's too easy to neglect the other tasks at hand once a long day is over. Lecturing can take a lot out of you- after a couple hours of lecture, you need a little time to unwind. Grading can be a little stressful too. All in all, it can be quite draining emotionally. I guess that's part of the process of learning how to be a professor.

It's not like I can pull entire weekends in the library to catch up- when you have kids, your hours are a bit limited. I was asked yesterday whether this is really worth it- what I'm doing- leaving a decent career, with kids, to go back to school and take the long road to getting a PhD. As my old boss at the PA told me, getting a PhD is "a younger man's task". It might take years to recover from this.

But with a little distance, and a little more sleep, I remind myself what the whole point of this was. It's not about next year, but the next 30 years. Demographics are on my side- a whole generation of faculty and administrators are retiring in the next decade, and not so many in my generation went into academia, so there'll be a huge shortage of tenured faculty. On the other hand, there's such a surplus of Web developers (at least in the NYC/NJ area) that I've joked about issuing hunting permits. (Though I'm not against the H1-B work visas for foreign programmers- I'm a free market advocate, after all.) Long term, it was better to leave all that. Short term, the market is still decent, though.

This will be our fourth New Year's back in Albany- the last, with the ups and downs of student life. This was never meant to become a permanent lifestyle, but now that I'm back on track, it's a straight run (if not a smooth or easy one) to graduation- one more very big project left- the dissertation to research and 2-400 pages to write about it. Allegedly it writes itself once the research is done. We'll see about that...

Marathon grading session

After digging out from the snowstorm all day, off and on, I'm finally able to resume grading, and getting ready to calculate final grades before tomorrow morning.

I think this is going to be a late night, at the rate I'm going. It's the hand-wringing that does me in when grading- when I'm grading decent-sounding papers that miss the point of the assignment. As I proceed down my little checklist (really, an Excel grid) awarding points to each category, I'm finding myself, over and over, having to skip entire columns. Do I just post an average? I need to invent another function aside form AVERAGE() and SUM() for these fairly decent papers that would get a 50% or less just based on the point system.

But for better or worse, it'll just be the point system for raw scores, followed by a curve. I - guess - that's - fair - somewhat. After next week, I'll have finished teaching my twelfth class in six semesters- which sounds like a lot until I consider that I'm booked for 6 more classes this Spring, as a full-time professor. But next semester will be different.

Really.

Done- almost.

I'm done.

I finished my paper last night following a 10+ hour session in the computer lab. I assembled all my disjointed notes, code snippets, and the odd XSLT template (which I know is redundant) into a 32-page paper which seems surprisingly coherent after re-reading it this morning.
So pending some bureaucratic paper shuffling, I'm finally, officially, a PhD candidate.

Somehow 32 pages in a day doesn't sound that impressive now that I'm now moving on to my dissertation- in fact it sounds like slacking off.

Now to finish grading for the semester-- I have, literally, hundreds of pages to go through in the next week.

Block!

I'm stuck.

Right now I have a half-foot-thick pile of papers to grade (or possibly more if I printed everything out), a half-foot-thick layer of snow over the driveway that needs to be shoveled (which I thought I'd shoveled this afternoon), final grades for roughly 80 students to calculate, and about 10 more pages to write for a paper. All due very, very soon. And I can't focus on anything anymore- I'll be amazed if this little blog post will even seem coherent in the morning.

I need a brain reboot. I think that used to be called sleep. Maybe in the morning, I can get wired on coffee and jump back into writing my paper- most likely in hour rotations between tasks, and after I send the kiddies off to school. Off to bed, then.

Falling into place

My GPS paper is finally falling into place. It's about context. It's about OLAP, in particular, that neat little way of extracting information out of data, though perhaps "neat", "little" and "OLAP" don't belong in the same web page. But it's yet another paper written out of some thoughts that occur to me while teaching. I've been talking about GPS, databases, and Business Intelligence a little in my classes this semester- all four of them- and it kind of dawned on me that Business Intelligence is the angle I need to work on, and the underlying Online Analytical Processing core, to make sense of the volume of raw data produced.

The point is this: transactional databases (OLTP) are commonly built to record data in some kind of efficient way. Most information systems are built to capture operational data. But the catch comes when you try to extract data in any form other than a simple query join. It slows down things. The more interesting the information you want, the harder the query, and the slower your database becomes.

Then, suddenly, you have to hack the database to become either faster at making queries, or else not do them at all. Usually, it's a combination, which results in a sloppy data model for capturing the operational data it was designed for. Then you end up with a really crappy database that doesn't really make anyone deliriously happy anymore.

So usually, the high level mining doesn't happen, because it compromises the operational support the thing was built for. The answer is to capture with one database, and export into another database built from the ground up for nothing but really, really hard queries that take a long time to run. It's run when you have to research something. But it's not on-demand, but staged, non-normalized, etc. But I'd say the overwhelming majority of developers don't know statistics or data modeling, and can't understand how to build an OLAP system.

Which brings us up to the failure of GPS to give us anything other than lat/long and what's nearby. I'm guessing we're years away from large deployments of spatially-aware OLAP systems, and whoever figures out how create a context for place will do very well for themselves.

Blackboard sucks

I thought WebCT sucked a little. Blackboard sucks even more. Granted, the application market space for web-based course management software isn't very big. The good products are actually Sakai and Moodle, both open source fare. Even if they weren't much better, the latter two products would at least have the virtue of being open source. Things that suck are made infinitely worse by being expensive as well.

Why does Blackboard suck so much? Because, just maybe, you can't sort information, like student grades, in any meaningful way. The web interface doesn't allow things like working side-by-side with another application, like the f-ing thing you're actually grading. It makes you print out a piece of paper so that you can type it into another application. So pre-1990's.

It gets worse. Entering a block of grades from another application can mean using a ruler to make sure you're entering text into the right row, unless you decide to edit each student's grades individually, for each grade, in a long list of exam scores. You'll have to use the scroll bar at the bottom of the browser window, because the name is in the first column, the little text box where you need to enter the number is in the last column, and inbetween are a list of links that don't say anything useful and don't lead anywhere, and yet force the table to occupy the full width of the computer monitor.

Course management software should give trend analysis. It should help analyze student performance. It should actually make things easier rather than harder. Instead I have to cut and paste into Excel to identify patterns. Can you track student issues, like manage rubrics as an entity with separately scored items that contribute into an assignment score? No, you get a single number to enter into an assignment, and a single HTML textarea to type your unformatted comments.

It would be nice if you could assign metadata to (i.e. tag) test questions, assignments, or even elements in a rubric and identify patters of errors- cover Chapter 6 in more detail, give an extra self-assessment, etc. Oh well...

But I hacked a worksheet in Excel that generates both the total score and a block of text with the feedback of what points were lost where, which I can cut and paste back into Blackboard. Or, I sometimes turn the Excel sheet into a database report, print out the comments, and hand them back individually. Which completely violates the purpose of such a tool, to help manage courses. Now it's overhead, because students expect to push a button and have Blackboard tell them what their GPA is going to look like, rather than track their own grades.

Making that work wastes many an hour over the course of the semester.

Which really sucks.

Seven down, two to go?

It's about the end of my seventh semester in the PhD program. And nearly the end of teaching my twelfth class. I think I'm a year away from graduating, though having just taken a full-time faculty position, I have a feeling that this is going to be one busy summer if that's ever going to happen.

This year has been an absolute blur. The summer seems to have lost a month. Somehow, a month vanished out of this semester as well. (I do have attendance logs that seem to demonstrate that at least I was teaching every day I was supposed to. I just don't remember it.)

I'm almost out of the woods as far as grading is concerned. Not much else to log into the system, though one class seems a little sparse- I have to comb through emails, papers, and the Dreaded Digital Dropbox for more stray assignments, that enormous black hole of jumbled, random uploaded files with such charms as homework assignments that turn out to be just half-page Word files explaining that the assignment couldn't be completed because of some reason or another.

And I left my USB key at school in one of the instructor podiums. And my laptop is suddenly dropping the letter "b" in at random whenever I use the arrow keys, though I hope that the last few rounds of compressed air around the keyboard may have finally cured that.

Ah, and I spent my 13th wedding anniversary teaching, in lab, grading, in office hours, and composing an exam, only to make it home shortly before my wife went to bed and I returned to my computer for more grading. It's the fourth time in a row that my anniversary fell during school finals week. But we got married in Grad school, and it was finals week then, too. Needless to say, she's quite sick of my schoolwork. But this is the last really hellish semester before graduating- having a job might lengthen the time to do the dissertation, but I'm not allowing it to get this bad again- this 24/7 lifestyle can't go on forever.