Find the thread

Of all the inconvenient times for my car's rampant hypochondria to manifest itself, the Sunday evening before an early-morning Monday jaunt to the airport is the worst.

Last time it was tyre pressure. The warning light came on, so I dutifully stopped to check everything out. The shortfall was smaller than the measurement interval afforded by the compressed air machine on a petrol station forecourt, but I managed to get enough air in to turn the light off. Coughing into the hose would probably have given enough pressure…

This time it was oil. The dipstick assured me there was plenty of oil, but I don't like warning lights, so I stopped at the first petrol station on the motorway at o'dark-hundred on a Monday morning to buy some engine oil. Of course they were out of the type of oil that the Beast's exacting tastes require, so I had to go on to the next place. This place did stock the right approved oil, so I was able to continue my journey fully lubed up.

This sort of thing is why I hate idiot lights. Give me a measurement! These days, with any number of digital displays on cars' dashboards, lack of real estate is no excuse. The Beast is actually pretty good about this, with physical gauges for both oil and water temps as well as turbo boost pressure. There is also a display for tyre pressure that can be put up in the central virtual gauge in front of the driver, between the speed and RPM.

What drove me nuts about my old car was the lack of these displays. On a turbo diesel, especially when cold, you don't want the turbocharger spooling straight away - but it's really hard to avoid it when the only instruments you have to measure boost pressure are RPM and engine noise!

Software designers on the other hand have a tendency to go the opposite way, with too much information being thrown at the user without context. The happy medium is to show the information, but include some context indicating what is good and expected, as opposed to what is out of the ordinary. Here car dashboards show the way: don't just have an idiot light that comes on when the driver is going "too fast". Instead, have nice big clear dials showing vehicle speed and RPM - but include red lines on both displays to indicate where the danger areas are.

Note: I’m not implying that all information should be displayed all the time. Software developers often fall into the trap of displaying every piece of information they can get their hands - well, digits, anyway - onto. The only result of this smorgasbord approach is to overwhelm users. This way, important things can easily get drowned in the noise. Operators then end up missing some thing important.

The most famous recent example of users drowning in data and missing the one really important piece of information was probably at Target. You may remember Target from such data breaches as The Largest Retail Hack in U.S. History - well, until Home Depot, at least.

The most interesting aspect to me, however, was that Target did in fact have systems in place to detect exactly the sort of activity that was involved in the breach. Those systems worked perfectly, and did indeed detect the breach in progress and alert operators. The operators simply missed the alarms.

How does that happen? Easily. At any given time there are scads of alerts flying around any sizeable IT environment. The trick is filtering out the all-important signal from the all-consuming noise - and this is where Target failed.

Target should not be blamed too much, though - most IT organisations are in exactly the same situation. The problem in IT used to be about too little information - but now it’s about too much.

And now is where I finally get to my point. My new gig is for a company called Moogsoft, which is working to solve precisely this problem. Our technology is able to sift automatically through masses of raw event data, figure out what is important, and show those important alerts to the people who can actually do something about it. The way we do that is with various algorithms, and I can geek out for quite some time on the information theory aspects of that - but the proof is in the results we are already bringing our customers.

Bottom line: information is good, but it has to be possible for users to consume it. Useful¹ context is critical for people to be able to make sense of data instead of simply being overwhelmed.

I finally had the chance to see the new Smart forfour in the metal - and my, it's a bland, ungainly thing.

Lest we forget, the original Smart car, retroactively named the "fortwo", was a design that embraced its tradeoffs and compromises.

Designs that have the courage of their convictions are the ones that stand the test of time. Compare the original Fiat Multipla with the bland face lifted version that replaced it.

Original: definitely challenging, but the designers were trying to do something new and different, and the aspect of the car reflected this. Three-plus-three seating is still unique today, and on top of that, the car still fits in a "traditional car" footprint. No wonder these were instantly popular as taxis - one more seat in a car with almost the same exterior dimensions!

The updated version toned down all the quirks, becoming so instantly forgettable that I'm surprised owners didn't forget they owned a car. I imagine queues of confused Multipla owners back at the Fiat dealership: "I could have sworn I owned a car, but I can't for the life of me remember what it looked like!".

People defend courageous designs; the original Multipla and the first Smart have their partisans, but nobody will stand up for their unworthy descendants. If you're going to do something, do it all the way. By trying to please everybody, you guarantee that you will excite nobody. And once you cede that high ground, it's a race to the bottom on price, and there will always be someone hungrier and more desperate.

Instead, define what you stand for, and stand behind that definition with everything you do. This is how you gain and keep customers, by doing something that you and your customers care about. If your design shows that you don't care, guess what? Your customers won't care either, and next time they'll buy a Hyundai or some other interchangeable "appliance" car.

Today at the Geneva International Motor Show, Apple announced CarPlay, with is the new name for "iOS in the car". It looks great!

They are pitching it as "a Smarter, Safer & More Fun Way to Use iPhone in the Car", and I love the concept. We upgrade our phones every year or two, but our cars much less frequently than that. In-car entertainment systems are limited by the automotive industry’s product cycles, so they are basically already obsolete (in consumer terms) by the time the car hits the showroom. Enabling cars to piggy-back on the smart, GPS-navigating, voice-recognising, music-playing computers that we already carry in our pockets can only be a good thing.

If you take a look at a video of CarPlay in action, though, I see one huge issue.

Touchscreens in cars are a terrible idea. Old-school controls gave drivers haptic feedback: if you turn the dial one notch, you feel the click, and you get the next radio station, or one increment of temperature, or whatever it is. This is something drivers can do completely blind, without taking their eyes off the road.

It’s been at least a decade since cars could accommodate physical controls for all of their functions, so multi-function control systems like iDrive, Comand or MMI were introduced to help with the plethora of systems and settings that modern cars require. All of these systems allow users to navigate hierarchical menu trees to control in-car entertainment, navigation, and vehicle settings.

These systems are still better than touchscreens, though, because at least the driver’s hand rests on the one control, and that control still has tactile increments to it. The haptic feedback from those clicks is enough for seasoned users to be able to navigate the menu tree with only occasional glances at the screen - very important at highway speeds.

Touchscreens are terrible for non-visual feedback. Users have no idea, short of looking at the screen, whether their action achieved the result they wanted or not.

Apple’s suggestion of using Siri is not much of a fix. I like Siri and use the feature a lot in my car to dictate e-mails or messages, but it depends entirely on network access. Out of major cities - you know, where I take my car - network access is often insufficient to use Siri with any reliability, so drivers will almost certainly need to use the touchscreen as well.

I really hope that CarPlay also works with steering-wheel mounted controls, as those allow control with an absolute minimum of interruption. If we could have audio feedback that did not require network access, using the existing Accessibility text-to-speech functionality in iOS, that would be perfect.

My new-to-me car has an utterly annoying feature, where it will recommend changing up as soon as the engine is not actually at risk of stalling. With a straight, empty road, I decided to indulge it and see what would happen.

Tooling along in fifth, the revs were well shy of the 2k mark (I run a turbodiesel) and sure enough, I had a little 6^ in the middle of the dash to encourage me to shift up. The interesting thing to me was that the fuel consumption in fifth was about 5 litres per 100 km, while in sixth it shot up to nearly 10 l/100km. I made sure that the throttle was steady and I wasn’t otherwise affecting the results, but it was completely reproducible. As soon as I shifted back to fifth, fuel consumption dropped, and the car started nagging me to shift up.

This sort of thing is why I am suspicious of the current rush to Big Data. Unless you are very careful and you are sure you understand exactly what is going on, it would be very easy to get caught in exactly the same sort of trap and make wrong decisions.

This is of course assuming that other factors don’t come into play, further fudging the issue. In the case of my car, I suspect that somebody decided that having a shift-up indicator with its associated ecological do-gooder associations was more important than making sure that its advice was appropriate.

The corporate world can easily fall victim to the sindrome of "We need to do something! This is something - let’s do this!". Nothing is more frustrating than to hear that scarce resources have been squandered on some sort of wild-goose chase, instead of applied in the pursuit of a fully thought-out goal.

Far more projects have been sunk by an excess of mis-applied resources than by a straight lack of resources. Make sure that before you do anything, you know the goal that you are trying to achieve. Next, double-check that what you plan to do will move towards achieving that goal. Finally, make sure there isn’t something more important that needs to happen first. Hacking away on a feature that will benefit one or two customers is worth doing, but not at the expense of delaying the next version or cutting features that most customers care about.

Building features just for their own sake is also a quick way to get yourself caught on one horn of the Innovator’s Dilemma. Extra capabilities add complexity to your product, so you need to be sure that you have enough benefit to make the complexity trade-off worth your users’ while.

The easiest way of all to fall into this trap is to follow the herd. "But all our competitors have a hosted version!" So what? Do your customers want it? If you offered it, would they even be able to use it? If they do want it, do they want it today, or is it enough for them to know that you are working on it and will have it ready by the time they are ready for it?

Also, does it come in red?

Image by James Forbes via Unsplash

I love my Beemer. Sure, it's only a 320d, so given the choice (and funds) I'd far prefer something with an M in the name. For instance I don't understand why my wife rolls her eyes when I point out the many ways in which the E60 M5 Touring is the perfect family car.

See? Perfect.

However, this is a company car, so I'm pretty lucky that it's not a misery-spec Fiat Stilo or some other sin against driving. It has sat-nav, Bluetooth, steering wheel controls, and so on. Being an E90 means it's starting to be a little bit dated in the details, e.g. the Bluetooth is not 2.1, so no music streaming. On the other hand there's an aux-in jack in the arm-rest, so it's not as if I have wires trailing everywhere.

The great advantage of a company car is of course that anything that goes wrong with it is Somebody Else's Problem. Even my accident, annoying though it was, didn't cost me a penny. However, the disadvantage is that the user doesn't get much input; in my case, the options were saloon or wagon, in any of three or four colours. Mine is a shark-skin grey which I think looks great, especially when a little dust gives it a matte look.

When I received the car I was therefore surprised to find that it was an automatic, as that was most definitely not a standard option. After checking the VIN, my fleet manager and I ascertained that it was the right car, just with the wrong spec, but BMW said I could keep it anyway. I was skeptical of the automatic, never having had good experiences with slush-boxes in the past, but off I trundled.

Misery on a stick

Once I had avoided killing myself or causing an accident by forgetting that an automatic transmission in D will go as soon as the driver takes his foot off the brake, without needing active application of the throttle, I got quite used to it, especially trickling through traffic on the Milan ring-road. The auto-box also made a good fist of snow on the roads in Milan, even on summer tyres (I hadn't yet had time to change them). Just roll s l o w l y off the brakes, don't touch the throttle, and it goes just fine.

The one thing that still really annoys me is the way it steadfastly fails to select the right gear. Left to its own devices, it tries to keep revs as low as it possibly can, which on a diesel is very low indeed. This is not ideal going into roundabouts or sharp corners, as it feels very lumbering and understeery without some revs to keep things under control. In frustration, you might be tempted to switch the transmission to DS, a sportier mode that holds gears for longer and down-shifts more readily. This does OK into the roundabout, but unless you're in a boy-racer mood, the way it then proceeds to hold second gear all the way to the next traffic light can get annoying.

So in the end you resort to swapping gears yourself, but then what's the point of an automatic transmission - apart from giving your left foot a break, that is? You still have to think about which gear you want to be in, and you still have to take your hand off the steering wheel to get into that gear, but now there's also an annoying... pause... before the gear actually engages, not to mention those fun times when you switch from D to DS and then into manual, resulting in all your passengers kangarooing into the air and swearing at you because you got three downshifts when you only wanted one.

Surprise!

This is a violation of the principle of least astonishment. This is a notion of UX (User eXperience) that basically states that the results of an action should match the user's reasonable expectations. It came to my attention as part of the iPhone mute switch brouhaha, but it makes a lot of instinctive sense. If I flub a gearshift in my manual car, because I didn't rev-match or misjudged the clutch's bite point, I have only myself to blame, except in particularly egregious cases. If on the other hand the auto transmission does something "helpful" resulting in an unexpected outcome, I become very annoyed with the automation.

To cut a long story short, my next car will have a proper manual again, unless there's a very good reason why - for instance, that I have finally talked SWMBO into letting me get an RS6 Avant or an M5 Touring…

Yay, I finally got my car back!

If you haven't heard the story already: On the 13th of Dec I was driving home on the Milan ring road. It was rush hour, so stop&go traffic. At a certain point traffic stopped again, I stopped, and so did the cars behind me. Then I saw a car in my rear-view mirror, skidding along between the lanes of traffic and not managing to stop. Somehow he missed several cars, but hit me hard enough to bounce me into the car in front. Nothing serious – no airbags – but all the crumple zones, well, crumpled.

This is where it gets weird: I checked on the car I had hit first, made sure they were OK, then while I was talking to the driver we saw the driver of the car that hit me with his two passengers walking towards us. We turned back to his wife, turned back again, and they were gone, leaving their running car abandoned in the middle of the Milan ring-road!

We naturally assumed the car was stolen and called the police, but when they arrived and ran the numbers it turned out to belong to a gypsy. These guys register hundreds of cars – quite literally; the policeman told me they found one guy with four hundred cars in his name! - and then anyone in the tribe can drive them. The guy who hit me may have been illegal, under age, without a licence, wanted for something else, drunk, high, or any or all of the above, so he ran off, but the car itself was legal and even insured.

It just meant a cold couple of hours going through police formalities at the roadside, then getting the car towed and picking up a rental (company car FTW!). What I was not expecting was for the repair process to drag out so long. While waiting for my own car, and driving various rentals, I have developed a Borat-esque attitude to gypsies…

As I said on Twitter:

After six weeks of rentals, my BMW 320d is going to feel like a Bugatti Zonda FFX Sport. Biturbo. Will miss the constant shift-up light tho!

— Dominic 🇪🇺🇺🇦🏳️‍🌈 (@dwellington) January 24, 2012

That was actually a false alarm, as it turned out to be close to seven weeks. However, the rest of that prediction was spot-on. I got in my Beemer, adjusted the seat how I like it (as low as it goes) and already it felt like being in a Formula 1 car. I pootled along icy back-streets to the ring-road, and half-throttle in the merge lane felt like Mach 1. I have to admit I was grinning ear to ear with the joy of getting my own car back!

Fortunately I was keeping an eye on my speed, as the garage appeared to have reset some of the electronics, including my audible speed warning which goes off at 110% of the motorway limit (about right, given the Beemer's "optimistic" tach). I didn't really exercise the suspension or tyres, as I was on calls for most of the drive back, but that's going to be a job for the next time I'm driving back from a customer visit with no particular place to be in a hurry.

The weeks of rentals do give me an opportunity to review the two cars I had as successive replacements. Here are reviews of the Ford Fiestaand the Hyundai ix20.

Spoiler: neither of them can hold a candle to the BMW 320d.

The Hyundai ix20. Wow. Where to begin?

This was my supposed "upgrade" once it became clear that the repairs to my Beemer would be taking longer than anticipated. Actually, I'd have preferred to keep the Fiesta since while the Hyundai was from a higher rental category and with a larger engine capacity, being a diesel engine in a much heavier car meant that the difference was if anything in the Fiesta's favour.

The looks were another defeat for the Korean newcomer, as they were initially generic but inoffensive, but rapidly revealed the touch of the bean-counters. Views were obstructed and dirt gathered disproportionately in door handles. Inside it was the same story: the buckle for the middle rear seat belt hung from the ceiling right in the middle of the rear-view mirror's field of view. All the surfaces were nasty to the touch, and despite this car having only twenty-five thousand kilometers on it, the steering-wheel coating had rubbed off in several places. Even superficially nice touches turned sour; for instance, I was pleasantly surprised to see a Volkswagen-esque blue glow from the instruments, but upon looking closer, found the blue to be simply painted on…

I compared this car to an Aiwa stereo, as while most of the functions were there, including Bluetooth and voice controls, they were all slap-dash, low-quality and annoying. NVH was abysmal, power was non-existent, and the radio did this weird thing where tuning into a pre-programmed station would give a few seconds of static which gradually faded into the actual radio station.

The worst aspect of this car was the handling. It takes a lot to scare me behind the wheel, but the first time I took a corner at speed in this car I was terrified. I had absolutely no idea what the front wheels were doing. This wasn't just Fiat-style lightness, this was lightness combined with a total absence of feel. More time with the car did not lead me to tune in to its handling, more the opposite. Driving in the snow was the worst, as I could only tell when the car had lost traction or was under-steering by looking out the window.

Verdict: it's better than no car at all, and if you don't care at all about cars, it's reasonably comfortable and well-equipped. However, if you care even a little, this is one to avoid. If it's given to you as a rental, haggle, bribe, seduce, or if all else fails, pay for an upgrade.

The Fiesta was the car I picked up on the day of my accident. Edging into my adrenaline hangover, I was glad to have any car, even a small and under-powered one, to replace my poor Beemer. However I have to admit that I warmed to the little Fiesta pretty quickly. Even in rental-spec it had nice features like steering-wheel-mounted controls for the radio, the engine was zippy and loved to rev, and with wheels at the four corners it cornered like a go-cart. Up at motorway speed the engine revved pretty high, but was never obtrusive, unlike the Fiat Punto I got last time I had a loaner. It did lack a sixth gear, which would have helped a lot with the motorway cruise.

In town the steering was precise, and while it could have given more feedback, I was never in any doubt about what the front wheels were doing. This is another area where the Fiesta was unlike the Punto, or indeed any recent Fiat. The clutch was light and responsive, and as long as I kept the revs above 2500 rpm there was plenty of torque available to take advantage of short-lived gaps in traffic. The engine really did like to rev, but somebody at Ford had decided to fit a "shift-up" light and program it to come on just shy of 2000rpm, or about 500rpm before the power band starts. I, of course, took this as a challenge to drive absolutely everywhere with the light blazing, including while parking. Yes, I am actually twelve years old.

The cabin was basic but well laid out, with plenty of storage, all the controls within easy reach, and radio controls right on the steering wheel. This last is something I believe should be standard, as these are some of the most frequently used controls in a typical drive. For many drivers in fact the radio controls are used more often than indicators (a pet peeve of mine).

I even succeeded in getting the tail to step out while late-braking on a wet roundabout, and the slide just felt so right. In true Troy Queef style, I simply catch it with a dab of oppo and I’m away.

The Ford Fiesta: approved.