“When things get extra turbulent, the silly question moves the conversation forward”
Robot cheerleaders fail to excite Steve on a visit to trade show CeBIT, but a plea for more open standards in IoT does set his pulse racing
If there’s a single measure to apply to modern businesses then this year is all about agility. Despite the fact that nobody has standardised the unit of agility, everybody thinks they can measure it and demonstrate what it means. When trying to make sense of this type of problem, I fall back on a simple technique known as derailing someone’s smooth sales presentation with a silly question.
You see, silly questions are much more powerful than sensible ones. Especially when someone agile has been through and whipped up a show, a market, an economy, or the whole world into a turbulent mess in their passing wake. When things get extra turbulent, the silly question moves the conversation forward, by reminding us that there really are no sensible anchors in R&D, other than the ones we make for ourselves.
The same goes for silly topics. One of my invitations for this year’s CeBIT show in Hannover made mention of “robot cheerleaders”. As neither robots, nor cheerleaders, form part of my natural bailiwick, I was more than happy to have a look – which is the reason for the hastily snatched screenie of my heartbeat ECG ( below).
Even without cheerleaders, I was mildly excited – mostly because the last time I saw this data it involved some translucent gel and a young chap nervously shaving my chest.
It’s difficult to get clear electrical data such as this without those precursors, so the device I was holding that looked like a gamer’s steering wheel controller was quite the breakthrough in wearable, touchable contact sensors. The cheerleaders I mention were part of a video showing two exceedingly deadpan gentlemen, both wearing fleshy-looking flexible biosensors; as they stood up, little robots in front of them stood up too. There was a lot of late 20th-century pop video surrealness about it all – and for me, the killer problem was the overreach.
Nothing in the video proved that the movements of the little plastic homunculi were triggered by the fleshy plastic things stuck on the humans’ heads. There wasn’t a choreographic link. The humans got up and sat down in a very Daft Punk way, careful and rigid – they behave more like toddlers, clambering on and off their feet.
Nobody – robot or human – ever made any moves I’d associate with a cheerleader. I’ve included a picture of the video screen, but I wouldn’t blame the editor if he threw it out: it was a bad picture in a bad situation. [Fear not, it’s on the opposite page – Ed.] Which brings me on to the more obvious robots, and indeed those that we’re being told to expect great things from in the near future. Cast your gaze to SoftBank’s Pepper, opposite, which was originally developed by Aldebaran ( softbank.jp/en/ robot). It’s looking at machines such as this that devalues the achievement of the guys producing the ECG electrodes, because Pepper
is meant to be programmed cute.
Cuteness isn’t on my radar. When I want to think about how a system works overall, I don’t add “cute” to my list of tests – but Pepper is all about the cutes. I found the various Peppers on the SoftBank stand in a series of strange postures, something that also seemed to rather badly affect the attending technical team. Pepper, you see, is a curious combination of attributes, with a surface a bit like a washing machine, big doe-like eyes, a wasp waist and hourglass figure, a huge touchscreen on its upper chest, and very humanlooking ribbed plastic hands. Oh, and it’s short. To my mind, Pepper is about the size of a seven-year-old girl.
I think I could have found this disturbing, had the Peppers been operational. There was much work-in-progress, with actual engineers doing stuff. I was naughty and sneaked a snap of the laptop that appeared to be responsible for Pepper programming.
It was a very strange display, especially sat alongside the dancing assembly-line arms of Epson’s stand, where cuteness and the accompanying exploitation of humanity’s Skinnerian response takes a total backseat against the speed-ofmovement, force-applied, jobs-done world of industrial robotics. Not that Epson’s team was therefore brutalist or uncaring: I asked about the power consumption of its huge paper recycling device (the PaperLab), and they were happy to inform me this was a staggering 6.5kW. Not great news for PaperLab fans, but the news was a lot better on the current-generation office-sized inkjets, which squirt on the ink that PaperLab removes at a much more liveable 140W power draw.
The absolute top mark for me on the “intrusive, transformative, personal” front in technology came during a short appointment pause, waiting to see Fujitsu’s Dr Joseph Reger. He and I are somewhat like the good and bad Jedi knights of the press Q&A session; there’s no question I can think up that he can’t answer. He got an easy ride from me that day, because while waiting for him I was suckered into trying out the 3D glasses developed by a spinoff/joint venture of Fujitsu, with a name so thoroughly Japanese that I now can’t read back the business card I was given.
And I really wanted to follow up, because these glasses use a full-colour laser to paint the virtual screen you see, directly on the back of your eyeball. As a result, the dangling lens in front of your eye is supertiny, only really needing to bend the laser’s beam as it scans across the presented image. And since it’s a laser, and at any one moment a single point of light, it doesn’t need to be corrected for any faults in your eyeball. I find most VR headsets a pain, because I first wore glasses at the age of six; these are revelatory. Take my glasses off and I can’t see ten inches – but I can see the virtual display of the Fujitsu glasses, in gloriously clear and focused HD.
The barriers between me and a perfectly tailored VR display are still difficult to overcome, however, because this system requires that you keep your eyes fairly still, and I already know from my frazzled optician and a professor at Moorfields Eye Hospital that I’m an “eye mover”, not a “head mover”. When this technology catches up with people like me, though, there’s a distinct chance I’ll be a near-instant convert. All it would take would be a highperformance camera, some additional processing of the visual field for reality augmentation, and I’d be completely sold on doing without cumbersome lenses, awkward contacts and the like.
Which isn’t where I expected to be taken with an investigation that began with the teasing proposition of robot cheerleaders. However, it’s a measure of how far the entire IT business has come in just the past few years. I can remember walking around earlier CeBITs and losing count of all the laptops and servers, even the server racks. This year has seen a change to a selection of machinery that uses computers to get stuff done, rather than being computers that can run anything you care to put on them. All that said, I think the claims being made for robotics are at least five years too soon, before we even starting to look
at the social implications.
Agility vs openness
I was sitting in the press pack seats on the Software AG stand when the bigwigs swung past. It’s normal to have at least Angela Merkel come by and say a few things. This time we had one of her ministers, who had a lot to say about the wonders of the software integration and services sector. I wasn’t there for the politicians – the media circus surrounding Ms Merkel this year was boosted by the presence of the Japanese prime minister, which made it even more difficult to stick to any appointment on the sprawling fairgrounds of Hannover Messe. I was there for the speeches of the people doing the work: there’s something about the presence of a
few world leaders that makes these chaps raise their game. Perhaps because that huge rugby scrum of mainstream media reporters and camera crews might sit down for a rest now and again, and the right soundbite is gold for those guys. This year’s surprise for me, apart from the abrupt shift from PC makers and devices, over to device makers who might occasionally use a PC, was a closing comment from Karl-Heinz Streibich, Software AG’s CEO.
Looking out across the hall of stands, he’d had time to understand something that I’ve been struggling with for some time when looking at the world of small, simple computers now intended to control or report on just one or two physical sensors: the vast array of the Internet of Things. My problem has long been that being able to miniaturise a complete PC leaves far too many holes open in the whole system– holes that we’re really only learning to plug now, as the risk level of unintended breaches or uses for all those tiny little generalpurpose computing devices rises.
I thought I’d understood what the industry would do about this during a visit to the IoT World Congress last year in Barcelona. There, the exhibits were all about diverse sets of sensors and components, jointed together in a noticeably proprietary and often not-even-Ethernet network, then brought together and made IoTfriendly by a gateway device, sitting between their proprietary world and a general-purpose WAN or LAN.
Herr Streibich doesn’t like this approach. He made an impassioned plea that IoT implementers should always go for the most open hardware they can find to populate their estate of sensors, cameras, connections and actuators. Open wins over proprietary, he said – a strange viewpoint from a business whose lifeblood is the proprietary nature of its software product portfolio.
So I waited until the Q&A and gave him a second go at this topic. As I understand the current state of play in IoT, the biggest successes are found in the one-trick ponies: businesses that can take advantage quickly, because their grasp of their own equipment and what that means for their IT is pretty much underwritten by proprietary control of the entire hardware and software stack. Lexmark is a good example of this. By owning all the printer firmware, it can control exactly how it works as a plain IoT equal-status internet member device, even though all it does with that lofty status is to ask for new toner cartridges. Surely, I asked, there has to be some proprietary-ness in the mix, to make such deployments really work?
His response confirmed something that I’ve been quietly thinking for some time. All the problems with IoT – even the daft ones that come up around gussied-up consumer devices – are failures in transparency. You can’t interrogate your poorly coded, insecure IoT fridge over your own network because it isn’t an open standard with a neat walkthrough somewhere on the web.
It’s pretty simple to be transparent in software, with debuggers and compilers and the like, but the R&D houses have been concentrating so hard on getting anything with an “IoT” badge on it out of the door, they’ve barely thought about how to allow the purchasers and operators of their kit to snoop around on the inside of the tiny computer they’ve co-opted to some incredibly boring job.
Take a look at the NAS box market. I think of NASes as the precursors of the IoT: they’re computers and they started out doing just one job – handing out files. It didn’t take long for that list of jobs to expand, and the software writers wasted no time in making it easy for the enthusiast or home user to get inside the box and take a look around.
It needn’t be all that easy – SSH shells and telnet access require jumping through a small number of hoops before access is allowed – but that’s okay too, as a way of keeping out the idle and destructive fiddler. Once inside and working with a command line, even a humble NAS can be made to do all sorts of useful things, such as adding on background file sync services, or getting the machine to work as a media file server as well as just a plain old file server. (Okay, so maybe I was being harsh, criticising IoT for only giving decent CPUs boring jobs to do; described this way, the life of a NAS box sounds pretty boring too.)
The point here is the open-ness. As Herr Streibich was insisting, you can see the advantage to being able to completely describe what that little lump of chips and LEDs in the corner of the room is up to, and the more proprietary it gets, the less likely it is to be successful. The next time you’re considering an apparently IoT CCTV system, or a flashily not-quite-IoT home heating controller, see what happens when you ask the vendor if the system is based on any recognised, open standards, and how open it is to you taking a good look around inside the tiny little OS and filesystem it comes with. If you’re told this isn’t needed or not normal, maybe make a different purchase. Openness and transparency are the only good defences we have.
“Being able to miniaturise a complete PC leaves far too many holes open in the whole system”