Dan Quayle's Secret Occupation

In Yahoo's translation service (formerly Alta Vista's Babel Fish), enter

tomate

and ask for translation from Portuguese to English. It responds with

tomatoe

Given that this problem does not occur when translating from, say, Spanish to English, it suggests that Babel Fish has a separate vocabulary for each pair of languages, rather than an intermediate semantic representation. This would also explain why the set of X-to-Y pairs is fixed.

Google Translate doesn't suffer from this problem. They allow a free choice of source and target languages. From what I can tell, the superior technology they employ to solve this translation problem is to avoid translating anything at all. (Seriously, whether it's their HTML parser or their core translation routine—I haven't invested time to investigate, and I suspect it's the former to blame—their “translator” routinely returns the input unchanged as output.)

My Slow Email Movement

I've never been a slow-anything person, other than riding slowly up mountains, and that's only because of the weakness of flesh, not any unwillingness of spirit. So it's not often that I embrace a Slow movement.

But on my Web page, I currently say

Every month I get over 10,000 messages. Of these, just over 8000 messages are spam. In this same time I send over 1000 messages. I am, in short, a full time email employee who gets to do a little teaching and research on the side. You know, as recreation.

If any of the deans or assistant deans or vice-deans or sub-deans or deans-in-waiting or deans-in-law at Brown are reading this: I'm kidding! Everyone else: I'm not!

For a day or two, I played with Google Mail on my mobile phone. Then, one day, I was lost during a bike ride, so I pulled out my mobile to find my whereabouts on Google Maps...and found myself checking my email. Soon after the apps ceased to work on my T-Mobile phone, and I was happy to not investigate why.

For the past year or so, I have rarely been checking my email when I travel. That is, I check it once every two to three days. And here's something amazing. If I wait a day, it takes me about an hour to restore my mailbox. If I wait two days, it takes me about an hour-and-a-half. If I wait three days, it still takes me about an hour-and-a-half.

These numbers are slightly misleading. They mask critical tasks that require real attention to detail and will take much longer than a minute to discharge. But those tasks are relatively few: I can be gone for two weeks and find only two or three such tasks lying in wait when I return. Which suggests I'm significantly promoting in importance things I do encounter daily.

There are other knock-on effects. You've played email ping-pong, right? Everyone treats their mailbox as a task-manager, so you get a task, you reply or forward to put the monkey on someone else's back, they do the same to put it back on yours, and suddenly you've lost an hour of the day (because studies haved, shown that these context-switches are extremely expensive, though as computer scientists, we should have known that). And, since you and your correspondent are both on-line, your reply begets their reply, and so forth. Congestion-control through exponential-backoff, anyone? (This is why I enjoy clearing out backlog during times when lots of people are on vacation: significantly fewer replies.)

The backoff strategy shows where our email user interfaces have gotten it wrong. They show us when we received email, but who cares; they should instead tell us when we should be replying to email. And that “when” should be a combination of when we need to (based on message content) and when it would be prudent to (based on correspondent habits).

So I'm making a conscious decision. I'm going to go Slow on E-Mail. I'm going to treat it as an addiction, like drinking too much coffee. There doesn't seem to be a simple, prescriptive or descriptive classification of addiction treatment akin to the seven stages of grief; much of the on-line material about treating addiction is rather disturbing and possibly borderline dangerous (and, mark me, there's a megachurch out there somewhere that is going to make a killing off faith-based treatments for email). So I'll have to figure this one out on my own.

Do feel free to drop me a note with your thoughts.

Just don't expect me to respond.

Gaunt from Charts

I've never really cared for Gantt charts. I've seen them but never really studied them, because they invariably represent the graphical representation of something entirely fake (such as time-lines of work in grant applications). Indeed, they've always seemed mildly worthy of suspicion. When 37signals got into some trouble for refusing to support them in Basecamp, that just confirmed my opinion that 37signals was a slightly eccentric but righteous organization.

Two weeks ago I was scheduling the dates when homeworks would be assigned and due in my course this semester (programming languages), and I wanted to check on the distribution of homeworks and ensure there were never more than two homeworks out concurrently. I set the dates in a calendar and tried to view multiple months at a time, but the result was somehow unsatisfying. I thought for a moment about how I'd like the information presented. What I really wanted, I thought, was a picture—a picture of the homeworks stacked as bars—

Oh wow. I wanted a Gantt chart. (As is so often the case, wisdom follows from necessity.)

Now nothing spells corporateness like a Gantt chart, and nothing spells corporateness like Microsoft, so I figured it to be an ideal match. (Besides, I already knew there was no point trying to do this via 37signals.) Creating a table of the dates in Excel was trivial, so all I needed was to find the Gantt chart mode....

This is how Microsoft wants you to create a Gantt chart (short of buying Microsoft Project, I guess). It's startling to think someone actually wrote those instructions with a straight face. For what it's worth, there's a fine YouTube video that explains this process interactively. Reading the comments is interesting: compared to the usual drivel that people post, just about everyone here is focused, on-topic...and damn grateful.

MIR3, This is Earth Calling

In the aftermath of the Virginia Tech massacre, campuses are creating emergency notification systems. Brown has outsourced this to a company called MIR3 Intelligent Notification. We had a test run today.

I was aware this morning that there was going to be a test. But I was out for much of the afternoon so I forgot all about it.

I was at my desk this afternoon working on something important when I got a call. It was clearly an automated message (initial pause, followed by a slightly robotic voice), and all it said was something along these lines: “This is an important call. Please press 1 for an important message.” (Certainly the second sentence was verbatim.)

That's it. No identifying information, nothing, nada.

If you've ever worked from home, you know that this is precisely the kind of message that you get mid-day from guys trying to sell you timeshares in Florida. Same kind of voice, same lack of identifying information, same pretend sense of urgency to con you into listening further.

So I hung up, quietly cursing that the damn telemarketers had somehow managed to get my office line.

Only an hour later did I realize what the call was about.

Today they also sent us emergency notification email messages. The messages came in the name of a Brown administrator but from the MIR3 domain name, and the headers had enough to trigger the suspicion of many a spam filter:

Date: Thu, 6 Sep 2007 12:34:00 -0700 (PDT)
From: Walter Hunter <343821_6286449@notify2.mir3.com>
Reply-To: MIR3 System <343821_6286449@notify2.mir3.com>
To: my Brown email address
Subject: Faculty/Staff Notification - Test Only issued at 9/6/07 12:30 PM

It's even more likely to be binned in my case, because I know and have corresponded many times with Walter Hunter, and never through this email address.

Unsatisfied by my unresponsiveness it sent me the same message again, eleven minutes later.

The message body also demonstrated good attention to detail:

We are attempting to verify the accuracy of our data base. Please press or select "1" if you are a staff member

How's that again? I keep pressing 1 at my keyboard...but my email program doesn't seem to know what to do about it. (Don't overlook the charming Victorian prose touch: “data base”.)

And then:

!!! You may respond by doing one of the following: !!!

Nice touch, the three-exclamatory-marks. If the headers made it through a spam filter, this should give the message a fighting chance of being trapped.

And finally, one of my response options:

* Reply to this email with the corresponding number to your response on the top line within the body of the email, e.g., 1 for indicating that you wish to use response option 1.

Option# Response:
1 Faculty or Staff Member
2 Contacted in Error

Clearly these folks are adherents to the rule that, in an emergency, you should make your sentences maximally complicated. The logic is obvious: that's how you test whether the recipient is still clear-headed or is already suffering from smoke-inhalation.

Of course, this is why we conduct tests: so we understand how well our systems work and can, in turn, improve them for when we actually need them. But this first run does not give me a lot of confidence in this company's knowledge of how to create trustworthy communication.

Now I'm waiting to hear now many millions we spent on these “professionals”.

Time Travel

On Windows, a file has at least three attributes: Created, Modified and Accessed. while Modified and Accessed could be in any order (it may have been modified after you last accessed it), the one invariant you would expect to hold is that Created is older than either of the other two. Here's a file I just saw on my file system:

Created: Wednesday, August 29, 2007, 9:45:20 AM
Modified: Friday, April 06, 2007, 11:34:55 AM
Accessed: Friday, August 31, 2007, 2:33:41 PM

After considerable searching I finally found a Microsoft knowledge base article about this that explains how such a thing might have happened and, even more usefully, an article that explains the true semantics of these names and articulates the issues clearly. This latter article shows just how subtle it is to define precisely what these times mean—and equally, why it is equally important to define these concepts precisely.