New elevator technology

I had a chance to visit 7 World Trade Center today, the newest office high rise to open in New York.

Instead of having up and down buttons outside the elevators, there’s a numeric keypad, where you key in the floor number you’re going to. Then an LED display tells you which elevator to wait for. Once you get in the elevator, you don’t have to press any buttons (and there are none to press).

This is more efficient than the old system, in which two people who were going to the same floor might have taken separate elevators, adding an unnecessary trip. Presumably, during the early morning rush, it is able to clump people going to nearby floors into the same elevator, thus getting more people to their destinations faster by intelligently optimizing elevator schedules on-the-fly, instead of letting any arbitrary person force any arbitrary elevator to take them to any arbitrary floor.

Can you guess the usability bug?

(select this paragraph for the answer:) People who aren’t used to the new system come into the lobby and see an elevator with an open door. They jump into it, and then get stuck going to some random floor because they can’t key in their destination once they’re inside. (end)

In that old Jack Lemmon / Shirley MacLaine movie The Apartment, a 1950’s-era accounting firm accomplishes the same trick by putting human operators (“elevator girls”) in each and every elevator, and having a human traffic director in the lobby direct people to the elevator that will go to their floor.


Zagreb / Edinburgh

On May 17th I’ll be speaking at Web.Start in Zagreb, Croatia. “Web.Start is the first regional conference catering to the authors of Web projects, aiming to help them realize or improve their projects and helping them reach the knowledge and contacts that will help hem position themselves on the market, fund their projects and achieve success.”


Inadvertently, I referred to the UK as “England.”

In my defense, I really did think that I was analyzing data on customers in England specifically. Later I realized that our order form dropdown only has a listing for the United Kingdom.

By way of apologies, if you’re in Edinburgh when I pass through there on May 19th, let me buy you a pint. Sign up on this wiki page.

Guest editorial in Inc.

Inc. (the magazine) is a great resource for entrepreneurs. For the last several years, I’ve been reading it cover-to-cover the day it arrives.

That’s why I’m incredibly honored that they invited me to write a guest editorial about recruiting and internships in this month’s issue. Thanks to professional editing, it feels a little bit polished compared to my usual style. I don’t think I would write, “Ah, college.” I do remember writing, “Get me a frosty cold orange juice, hand-squeezed, and make it snappy!”

The picture that accompanies the article in the printed version shows last summer’s interns at Top of the Rock:

VBA for Macintosh goes away

The first few versions of Excel (1.0 through 4.0) had a rudimentary macro programming capability using a programming language so embarassing that it never had a name, although it was sometimes called XLM (its file extension).

In 1991, Visual Basic 1.0 had just shipped to rave reviews. Combining a graphical UI builder similar to NeXTSTEP’s Interface Builder with a simple Basic programming language that was highly compatible with QuickBasic, it rapidly became the best selling programming language, a position it maintained until droves of developers switched to web development.

Much as professional programmers sneer at the Basic programming language, market research unambiguously showed that about 2/3rds of the kinds of accidental programmers who develop macros preferred Basic to other languages and perceived it to be easy.

Thus, the obvious choice for Excel’s next macro language was some version of Visual Basic.

There were a bunch of complicated requirements, though. Excel was cross platform. The Mac version sold very well. To be a good Mac player, Microsoft had to support Apple’s cross-application scripting architecture, Apple Events. Rather than implement two object models, the Excel team concluded that the object model for Excel had to be Apple Events compatible. For complicated reasons, the Visual Basic engine wasn’t object oriented “enough.” In particular, objects could have properties, but those properties couldn’t, themselves, be objects. Visual Basic 1.0 didn’t support things like “rows(1).cells(2).value” because the row object couldn’t contain another object.

The VB team implemented an all-new version, for both Macintosh (System 7 on Motorola 68k) and Windows (3.0 on 16 bit processors). This became Visual Basic for Applications, and, soon thereafter, the standalone version, Visual Basic 4.0.

The whole effort took quite a bit of work. However, it was seen as extremely “strategic.” Here’s what that meant. Microsoft thought that if people wrote lots and lots of VBA code, they would be locked in to Microsoft Office. They thought that no matter how hard their competitors tried (in those days, they were Borland, Lotus, and, to a far lesser extent, Claris), they would not be able to emulate the VBA programming environment and the gigantic Excel object model perfectly. At some point, any Excel VBA macro they tried to run would get in trouble and crash. This is the same reason apps under Mono, Wine, etc. hardly ever work the first time out of the box: in any large API or programming interface, there are so many subtle, undocumented details of the behavior, which programmers may be depending on without even realizing it, that any emulation environment will inevitably be imperfect. In the brittle world of programming, such imperfections often mean your program crashes long before it does anything useful. You don’t get partial credit when you try to emulate an API.

In essence, in addition to giving Excel users a nice programming language, Microsoft was building a highly strategic barrier to entry, and locking in Excel users, especially corporate users who are most likely to build large systems based on macros.

Eventually, all the Office apps came along: Word, Project, Access, Outlook. What was a strategic lock-in for Excel grew to have major strategic value for the whole Office system.

Last August Microsoft decided to drop VBA from the Macintosh versions of Office. Despite complicated technical explanations, every development decision like this is based on a cost/benefit analysis. Mac users are less likely than Windows users to have business-critical macros, simply because Macs are rarer in large business.

But what’s really interesting about this story is how Microsoft has managed to hoist itself by its own petard. By locking in users and then not supporting their own lock-in features, they’re effectively making it very hard for many Mac Office 2004 users to upgrade to Office 2008, forcing a lot of their customers to reevaluate which desktop applications to use. It’s the same story with VB 6 and VB.Net, and it’s the same story with Windows XP and Vista. When Microsoft lost the backwards-compatibility religion that had served them so well in the past, they threatened three of their most important businesses (Office, Windows, and Basic), businesses which are highly dependent on upgrade revenues.

PS: in researching this article, I tried to open some of my notes which were written in an old version of Word for Windows. Word 2007 refused to open them for “security” reasons and pointed me on a wild-goose chase of knowledge base articles describing obscure registry settings I would have to set to open old files. It is extremely frustrating how much you have to run in place just to keep where you were before with Microsoft’s products, where every recent release requires hacks, workarounds, and patches just to get to where you were before. I have started recommending to my friends that they stick with Windows XP, even on new computers, because the few new features on Vista just don’t justify the compatibility problems.

PPS: I was a member of the Excel Program Management team from 1991-1993, where I wrote the spec for VBA for Excel.

Deriving your demand curve using exchange rate fluctuations

When I wrote that article about how to set prices for software, I generally concluded that in many ways you were completely doomed:

“The more you learn about pricing, the less you seem to know… I’ve been nattering on about this topic for well over 5000 words and I don’t really feel like we’re getting anywhere.”

In particular, to set prices well, you need to be able to plot your customers’ demand curves, and it’s almost impossible to figure out what your demand curve is, because it’s so hard to charge different customers different amounts and get any kind of reliable data.

Sometimes, though, you luck out.

If you’ve been selling a product priced in US dollars to customers in Europe, you might actually have a bit of useful data. You see, the US dollar has dropped a lot in the past year. As the dollar falls, your product has become cheaper and cheaper for Europeans.

I looked back on the last year of FogBugz data, dividing the price by the pound sterling exchange rate, and discovered that our single-user license have fluctuated between 64 and 74 pounds, while our ten packs have fluctuated between 49 and 56 pounds, approximately.

That gives me just enough data to plot a segment of the demand curve for English UK customers.

The data is not very conclusive, but it does support some things that I might have believed anyway:

  • Large customers (on the left, enjoying the volume discount) are not price sensitive. There does not appear to be any correlation between price and sales. These are larger businesses that are used to spending money on things, and it’s not their money, anyway.
  • Small customers (on the right) are price sensitive. They’re startups and small ISVs that are used to watching their budget. They exhibit a classic downward-sloping demand curve as economics would predict, and they like prices that end in 9 a bit better than prices around it (there’s a small bump below £70).

On this curve, demand is measured in units purchased per day in England the UK. I’ve left the Y-axis unlabeled because it’s confidential sales data, but the shape is accurate.

There are a lot of reasons to be skeptical about this data:

  • Since the dollar has basically been marching downwards, lower prices (in £) correspond to later dates. I’ve tried to adjust for this but there could be any number of other reasons why sales are increasing in England the United Kingdom.
  • The data are only based on sales in England the UK, where different economic factors may be in effect that don’t apply to US customers.
  • Many of our customers may not be looking at the actual exchange rate, but estimating based on their imperfect approximate knowledge of exchange rates (just dividing by 2, for example, is probably common), or thinking in $.

Still, you may find this a useful technique to learn something about the demand curve for your product.


Outlook 2007: downgrade no longer

Search used to be horrible in Outlook. It was so bad that for all intents and purposes, you couldn’t search your old email. Instead, you were encouraged to carefully sort it out into a hierarchy of folders (shudder).

A bunch of third party fixes appeared. My favorite was called Lookout. It provided blazingly fast full text search. Searches took less than a second and really found things. It was built on DotLucene (now called Lucene.Net) an excellent open source search engine.

Microsoft did the only thing that made sense: they bought Lookout (the company) and took the product off the market.

People complained.

Microsoft finally put Lookout back up for download, but they sure weren’t happy about it. When Outlook 2007 runs, it checks to see if Lookout was running and disabled it if it was.

Theoretically, Outlook 2007 has search built in, although it’s not really built-in: it’s built on top of Windows Desktop Search, which comes with Vista and is available as a free download for XP.

The trouble is, Windows Desktop Search is just not that fast. When I “upgraded” from Outlook XP to Outlook 2007, the only new “feature” I noticed was that full text searches started taking about 30 seconds. About 100 times longer than they did with Lookout. And I couldn’t install Lookout: of all the Outlook add-ins in existence, Outlook specifically refused to run Lookout.

The only possible explanation is that someone on the Outlook team is getting paid a bonus for convincing people to switch to Gmail.

The story has a happy ending. Last week Microsoft released a patch for Outlook 2007 which fixed the problem for me (I have a lot of big PST files, which, I’m told, is why search was so slow for me). Now I can search old email quickly enough that I don’t forget what I was searching for by the time the results come up. It’s not quite as fast as Lookout used to be, but it’s a big improvement and makes Outlook less of a downgrade.

Moveable walls

Among other things, this week I’ve been working on the new office design with our architect, Roy Leone [flash site].

One of the problems with our cool office here on 8th Avenue is that we outgrew it too quickly. The construction of the second half of the office cost us $100 per square foot of which the landlord only paid $30. All of that money went into walls, desks, bookshelves, and a fishtank which we’ll use for about two years and then abandon. There’s a good chance that the next tenant will have their own ideas about workspace and demolish everything we did. All in all, it’s a big waste of money and environmentally unsound.

Worse, when you spend money on construction like this, it’s not really deductable. Here’s an example. Let’s say your revenue is $100. Your expenses were $80, leaving you with a profit of $20.

Now, let’s say you spend that $20 on office construction (what the IRS calls “leasehold improvements.”) It’s not deductible yet. You have $0 in the bank, but the IRS thinks you made a $20 profit. Now you have to pay the corporate income tax (35% federal plus 11% or so for New York) on the $20 even though you don’t have the $20.

[Technicality: when you move out, you can deduct it all, and you can depreciate it at a rate of approximately 1/39th the amount per year, so you could really deduct fifty cents of the $20.]

We’re going to need a much bigger space now: on the order of 15,000 square feet. To build that much office space could cost a couple of million dollars. With the lack of deductability, your bank account goes down by three million dollars. The landlord will pay a fraction of that, but not enough to make it affordable.

There’s a loophole. Office furniture can be depreciated much faster than leasehold improvements, over 7 years. So for $20 of office furniture you can deduct about $3 a year: better than nothing. Even better, office furniture is a real asset, so you can lease it. Now you’re not out any cash, just a convenient monthly payment, which is 100% deductable.

This is why companies build cubicle farms instead of walls, even though the dollar cost is comparable.

I think I’ve got a good workaround. We’re looking into manufactured walls. They’re reusable, recyclable, environmentally friendly, you can take them with you when you move, and you can finance the purchase. A reader who works for Haworth emailed me to suggest their Enclose walls system (illustration at right). I visited their New York showroom and they look great, so we’re working on figuring out a totally standard set of parts we can order from Haworth for each private office to create the same extremely nice diagonal offices that we’ve built in the past out of drywall, metal studs, and acrylic.

[PS Please do not take tax advice from me. I am morbidly unqualified to give tax advice. Ask your tax lawyer or CPA.]

Monday Morning Links

Tyler found out that the Grand Canyon Skywalk is a ripoff.

Gerry Gaffney interviews me about customer service for UXpod, the User Experience Podcast (23 minute MP3).

I just got back from a trip to Israel, having, in a fit of stupidity, decided that it wouldn’t be such a bad thing to take Air France and change planes at CDG. Nothing surprises the ground crew at CDG more than the arrival of the daily, scheduled flight from Kennedy. Every single thing about that airport is broken and it’s the worst place in the world to change planes.