Business of Software Blog

Video: robotic dogs playing football

Tweet

A cool video of Aibo dogs playing soccer, as part of the Robocup competition. The ultimate aim is produce a robotic team of humanoid robots that can beat the human world soccer team:

Tweet

Subscribe to the RSS feed

Fred Brooks and his omelettes: is your software un oeuf?

Tweet

In The Mythical Man Month, Fred Brooks likened shipping software in time with customers' expectations to cooking an omelette:

An omelette, promised in two minutes, may appear to be progressing nicely. But when it has not set in two minutes, the customer has two choices--wait or eat it raw. Software customers have had the same choices.

The cook has another choice; he can turn up the heat. The result is often an omelette nothing can save--burned in one part, raw in another.

I'm going to stretch, and maybe strain, the analogy. Sometimes you're about to serve the omelette and you realise that something's wrong. It's not seasoned right, or you've forgotten to put the peppers in. Or perhaps your customer actually ordered a cheeseburger.

So what do you do?

You can serve it up and hope the customer won't complain, at least not loudly enough for others to hear.

You can insist that it's fine: you're the chef, you've been cooking omelettes for twenty years and the customer wouldn't know a decent omelette from his elbow.

You can cover it with a large sprig of parsley to make it look nice.

Or maybe you follow it up quickly with an omelette 2.0 which corrects the deficiencies in the first omelette.

Or, possibly hardest of all, you accept that the omelette isn't right, slide it into the bin, and start again. That's hard to do though. You'll disappoint your customer, it means acknowledging that your first omelette wasn't up to scratch, and it can be expensive in both time and money. But sometimes it's the right thing to do.

Tweet

Subscribe to the RSS feed

Major product upgrades - is your current version good enough?

Tweet

Releasing new products is hard. Releasing major upgrades can be harder, but for different reasons. On the one hand, you want to persuade your current customers to upgrade; on the other you don't want to strongarm them, bleed them for cash or leave them resenting you. From your perspective, supporting older versions of products is time consuming, but from your customers' perspective there are often diminishing return to upgrades.

The more mature a product is, the harder, the riskier and the more expensive it is for customers to justify an n+1 product upgrade. And from your perspective, your n+1 version product has stiff competition: your version n.

This is truer for some products than others. It's relatively easy for you to try out, and then upgrade to, the next version of SQL Compare, say. It's a third party tool that you can evaluate with little impact, and uninstall if necessary. If you judge the benefits outweigh the few minutes of evaluation time and the upgrade cost of a couple of hundred dollars then you buy it. If you don't, then you don't. Easy.

That's not true for other products though. The hardest are mature, server-based, enterprise-wide, critical systems. If you're a SQL Server customer, for example, then you've got a difficult decision ahead of you. Should you upgrade to SQL Server 2008 when it's released? The upgrade costs could be enormous, so the benefits need to be huge. You'll need to retest your current applications, evaluate the software and plan a roll-out. And that's before you even consider the direct costs of upgrading. At up to $25,000 per processor those can be significant.

Two and half years after the release of SQL Server 2005, our stats show that only around 50% of people have upgraded from SQL Server 2000, a platform that is now eight years old. So what will the adoption of SQL Server 2008 look like?

According to SQL Server Magazine, somewhere between 10 and 15% of people will move to SQL Server 2008 within about a year of its release. Somewhere between 60 and 70% have no plans to move to SQL Server 2008. Back in 2006, the top reason for not upgrading to SQL Server 2005 was because there was no compelling business reason to upgrade. If that was true for the SQL Server 2005 upgrade back then, it's doubly true for the move to SQL Server 2008 now.

There's another reason I think uptake will be low. A lot of us have been stung by the latest versions of Microsoft Office and Windows Vista. In Sketching User Experiences, Bill Buxton uses the example of the typewriter keyboard. Say you're a keyboard manufacturer and you find a way of re-arranging the keys of the standard keyboard. Usability tests show it's easier for beginners to get to grip with, and it ultimately leads to a 10% increase in typing speeds, both in novices and experts. You still wouldn't release the new keyboard: the several billion people who are comfortable with the current, inefficient, keyboard would find it irritating, confusing and slow to use. They would never upgrade. I put Office 2007 and Vista into this category, with their arbitrary new ways of doing standard tasks.

Although SQL Server 2008 has no equivalent to the productivity-destroying ribbon bar in Office 2007, I do think that Microsoft's recent history with product upgrades will make people wary and skeptical. It certainly won't push them to upgrade quickly.

On balance, people will upgrade to SQL Server 2008 very slowly. Clearly, that isn't in Microsoft's interests though: they, logically, will be trying to persuade people to move from previous versions. In particular, they'll be trying to persuade people to leapfrog SQL Server 2005 and move straight from SQL Server 2000 to SQL Server 2008.

One strategy they might use is to withdraw active support for SQL Server 2005. SQL Server 2005's second service pack was released about a year ago. Microsoft currently have no plans to release more service packs for SQL Server 2005. This  might make short-term sense for them, but it's not the right thing to do for their customers. If I'm right, and the uptake of SQL Server 2008 is slow, then Microsoft should be looking after customers who are unwilling, or unable, to upgrade immediately. People are going to be using SQL Server 2005 for another five years at least, and they need to be supported.

I think Microsoft are persuadable - there's a poll running on the Microsoft connect site where they're gauging feedback. If you use SQL Server and don't have plans to upgrade immediately to SQL Server 2008 then I urge you to make your voice heard and push for SQL Server 2005 service pack 3. Here's the link:

https://connect.microsoft.com/SQLServer/feedback/ViewFeedback.aspx?FeedbackID=326575

Microsoft will eventually do the right thing. They'll have to. Either they'll do it now, or they will backtrack and do it later, much like they've had to extend Window XP's life. They don't even have to write a single line of code or test a single function for now. All they need to do is commit to doing it, some time in the future, to keep their customers happy.

Go on Microsoft - keep us happy.

Tweet

Subscribe to the RSS feed

Haptic interfaces - did the earth move for you?

Tweet

Last October I visited the San Jose Tech Museum with some other people from Red Gate. We stumbled across a mechanical arm-wrestling machine. Here's the twist though: the mechanical arm that you wrestle is connected with other, similar, machines across the USA. There's a webcam too, so you can see who you're wrestling. In California, Theo, Ross and I all cheered on as Anna slammed the arm of some jock over in New York against the table.

This is an example of a haptic interface: you interact with the device via touch, pressure and movement, rather than sight or sound. It's an interesting example because it's not just about you, the user, interacting with a device. It allows two people, separated by a large distance, to interact as if in the same place.

The arm-wrestling machine is a crude example of how physical gestures can be transmitted over large distances. Wouldn't it be cool if you could transfer more complex gestures?

Well, it turns out you can. Almost. There's a company in London called CuteCircuit who have prototyped what they call 'The Hug Shirt'. The idea is that you buy two shirts: one for you, and one for a friend. You both put on your hug shirts. Your friend then travels off, to California, or Scotland, or wherever. You think your friend might be lonely and you want to send her something beyond an e-mail or an instant message. You hug yourself, squeezing your hug shirt. The shirt digitizes the exact form of your hug and sends the data to your mobile phone via a bluetooth connection. You text it to your friend. Your friend receives a message saying "Bob just sent you a hug. Do you want to feel it?". Her phone sends the hug to her hug shirt, and the hug shirt hugs her. Your hug - its strength, its pressure distribution, the warmth of your skin and your heartbeat - is replayed exactly the way you created it.

There are other people working on similar projects. Ben Hui at Cambridge University has a plan to send hand-squeezes via mobile phones. This is similar to a project that researchers at MIT's European palpable machines research group were working on before the lab was shut down. Rather than a shirt, this could be built into mobile phones themselves. You could squeeze the phone and your friend would feel it, in some form, at the other end.

The value of these haptic devices is based on the idea that physical touch is important to human interactions. By transferring and replicating the physical touch you are, in fact, transferring and replicating the emotion.

I'm not sure that this is true though. The emotion you feel depends on more than just what you're sensing. It also depends on the situation that you're in: what your physical environment is, your mental state and your expectations. Here's an illustration.

Also at the San Jose Tech Museum, there's an earthquake simulator. You stand on a platform and the simulator replays the seismic data from an earthquake that's actually happened. We re-experienced the earthquake that hit Turkey in 1999. In the real thing, 14,000 people died. In our simulation, the platform juddered for a bit and then we walked off, a bit disappointed.

Two days later, I was eating at The Grill at the Fairmont Hotel, just opposite the Tech Museum. At 8:04pm, the whole room starting shifting. It jumped left, right, forwards and backwards. The stack of wine bottles by the wall teetered, but didn't fall. This real earthquake, an order of magnitude tamer than the simulated one, only lasted a few seconds, but it scared the shit out of me. Because of its unexpectedness, my mental state, its context, and its reality, it provoked emotions of fear, awe and wonder that the simulation didn't and, probably, no simulation ever could.

So will the hapticon kill the emoticon? I'd like to think so, but there's still life in the smiley face for now.

Tweet

Subscribe to the RSS feed

Apply seat of trousers to surface of chair - how not to write great software

Tweet

When PG Wodehouse was asked what the secret of writing was, he replied that it's to apply the seat of your trousers to the surface of your chair.

In software, the secret is the opposite. It's to remove the seat of your trousers from the surface of your chair. At least, remove the tips of your fingers from the surface of your keyboard. Go out and talk to users; get a pencil and paper and sketch; take a long walk and think. But don't code.

Tweet

Subscribe to the RSS feed