In conversation - the Introduction

We decided we’d talk about each chapter in turn. In this first instalment, we talk about the introduction. This time, we’ve transcribed the audio by hand.

Hakim: Well, I guess if we look at our draft Table of Contents, we could go through it and just talk a little about each chapter, and what we think we’re going to be writing in it. We’ll start pretty free form, and if we start veering off, then that’s fine.

The introduction does have a few interesting things. Who should read that book and so on is fairly straight-forward.

Adrian: Almost boilerplatey (he says, and then we get to it and it’ll be really hard.)

Well, we’ll write that bit last.

Hakim: So designing for humans what does that mean?

Adrian: (laughs) Well, I think the way that I pitched it in the talk that I gave at IoT London in December is that the people who’ve got a headstart in building the IoT are all the people who’ve already been deploying big sensor networks. People like IBM but also other smaller people who’ve been sensor networks and similar. There are a lot of mobile phone masts over the country, and they’ll all have telemetry systems that relay whether they’re on, what the weather’s like, all kinds of status information. There must be a lot of data to aggregate. And in industry, people are using RFID tags, and you can drive a lorry through a gate with a high-tech multitag recognizing RFID reader, and it can scan everything on the lorry without it having to stop.

And all the stuff that’s driving that is businesses looking for ways to improve efficiency. so they don’t have to send a man to check every mobile phone mast, because the mast itself has reported a fault and they can just look at that one. So a lot of it is efficiency driven. And because they’ve already built these networks they’re really well placed, and they can say “that’s cool: this stuff is starting to get cheap enough”. So it’s not just big companies that have to invest millions in a big sensor network, you can start thinking about giving them to people at home. So your fridge at home might realise it’s stopped working, and could send them a message. So some of this stuff is useful, but the overriding stuff that you’re optimizing for is efficiency.

So people keep talking about the Internet Fridge. And it’s going to be So Amazing, because your milk will have an RFID tag on it and the fridge will’ve read that and know what milk’s in the fridge, and when its expiry date is, and will know that it’s expired, and order some more for you, so it’s all automated. There’s no human involved in any of that. You’ve outsourced all of your milk consumption, except the bit where you put it in your mouth, to machines. So you get to carry the milk from your door to the fridge, when the Tesco man turns up with your weekly shop, that the fridge has ordered for you.

Hakim: So is this a problem? Is it something we don’t want? Is it somethign we don’t want to talk about in the book?

Adrian: Well, it’s something I don’t want.. I don’t want that loss of agency. Suddenly on one level I rail against Best Before dates in general. It’s what a lawyer says is a best before date, and it’s really wasteful. You end up setting these rules that sometimes the milk will have gone off in X days. Unless the fridge becomes even more intelligenct, and can check if the milk has gone off… It’s all about packaging and consumer things, and giving up control.

But there are times when it would be useful: there was an example in the Smart Furniture Manifesto, and he’s talking about a trash can that you can buy from Amazon that scans barcodes. You can scan them as you put in the bin and later access the list of things you’ve thrown out, and decide if you want to order them. It’s almost the same as the fridge, but you make the decision about whether you want to order new ones. And there are some issues, because it’s at the point that you’re throwing the thing out, and if you’re throwing out a milk bottle because you’ve run out, then there’d be a delay before you get a new bottle. But you’re then actively deciding that you want more milk, rather than the fridge magically deciding to buy new milk.

Hakim: But in the old days, milk would magically appear on your doorstep in glass bottles…

Adrian: I have some great photos of some cheese that was delivered on my doorstep, when I’d been away for the weekend in Germany… it had been really hot, and the the milk had separated into curds and whey.

{parenthesis about discovery of cheese}

Hakim: you were talking about the M2M (Machine-to-Machine) big business IoT stuff, and funnily enough, when I looked at Wiley, our glorious publishers, they already had a number of books on IoT, and a lot, though not all of them, seemed to be focused on this enterprisey definition of the field. And it does strike me that by saying “Design for humans first”, we’re not saying that all of IoT is about that, but perhaps that our book is first and foremost about that? Has a different focus to thse other books.

Adrian: I think so, because that’s where I’m coming from at it. It’s not that there isn’t a place for this optimizing and efficiency. An example from my talk was a project that I did, sticking a load of sensors in a chicken silo. That was an industrial process, with the sensors that no human would ever have to interact with directly, so it was exactly about optimizing for efficiency in that scenario. But in the scenario where I’m involved with it, I’d like my life to be enriched by it, not just making me more efficient.

Like labour saving devices, you fill your kitchen with devices that give you more free time, supposedly, and then you spend that time working anyway, so you haven’t really gained any free time.

Hakim: There was a wonderful paragraph in Martin Amis’s London Fields, I think, about a character who used to love cooking, until he’d bought so many gadgets to do the cooking for him that he didn’t enjoy it anymore. So it’s a known problem with technology.

Adrian: I do have a bread machine, and I do use it all the time. And I wouldn’t make bread if I didn’t have it. Though I do make pasta, and cakes by hand. When I make cakes I don’t use a mixer. Just a bowl and a wooden spoon. And I don’t know why, obviously I’m just a bit of a luddite.

Hakim: There is a remarkable thing that technologists tend to have their luddite spots. Like a lot of geeks love to write in a moleskine…

Adrian: That’s interesting: what is it about a moleskine, or paper in general? There’s a different quality to writing on paper than on a flat plane of glass on your iPhone. If you approach things that designing for efficiency is just good, you end up with a very mechanistic world, and is that a good thing? I think we should be trying to make a world that’s more fun and playful. A better analogy, as IoT starts intersecting with humans more, isn’t to make things more efficient, because that’s not what we want, but to make things more magical. And magic is quite a good metaphor for it, because you start to get “enchanted objects”. So you get things like the WhereDial, which isn’t necessarily particularly efficent, because it’s only in one place, and it’s not portable, and when you update it, it’s not especially efficient to update.

Hakim: Well, with the WhereDial, the closest thing I can think of with that and the Weasly clock, and the Google Latitude DoorBell remind me of, in Finnish Mythology, one of the heroes, Lemminkäinen leaves a comb with his mother, and when he’s hurt, the comb starts bleeding, and his mother and wife know he’s in danger. And I think the WhereDial gives you that. Only without the blood.

Adrian: You could maybe build one that did bleed. And maybe that would be a more interesting way of doing sensors. There are lots of IoT things in healthcare. Things like GlowCaps, where if you haven’t taken pills then the caps glow. And after a certain amount of time, they can ever trigger a message to your doctor, or a loved one that you haven’t taken medication. And we’ll get onto that in the Ethics chapter… is it still Big Brother if it’s not the Government that’s watching you but your daughter, say, if you’re an old person in a home.

Hakim: well, presumably, you would decide…?

Adrian: well somebody decides that, but who. “Yes dad, you’ve got to start using this GlowCaps thing”. There are definite ethics and etiquette issues. We get that anyway, because loved ones would call you, but monitoring your health and alerting has issues. And at the moment, it sends you a text message, but maybe the alert could be a comb that bled… It would only do it a bit at first, so you might not notice, and then more insistently, the unhealthier you got. And maybe that would be a nicer way to deal wth that.

Hakim: (laughs) well, probably not actually.

Adrian: No, but the fact that you can think about it, not from “what is the most efficient way to alert…” OK, so if someone is having a heart attack then you do want the most efficient alert. But in many cases, you want a more ambient way of communicating.

Hakim: I’m almost wondering if the subtitle of the book could be “Designing the IoT for humans”. It seems to be a big theme for us.

Adrian: A whole new direction. We should have sections on whether things are “glanceable” and so on. But we do need to decide whether, if we’re going to fill our lives with billions of tiny computers, we don’t want to be spending all of our time seeing notifications popping up to say that so-and-so has just checked in 3 miles away. These things are already too annoying, just on my laptop screen, but I can close that and walk away. But if my whole house is telling me that, forcing this info down my throat then…

Hakim: I tend to turn all those notifications off, in Skype and so on…

Adrian: I haven’t, because it involves going out and finding out how to do it… but the defaults need to be sensible.

Hakim: And that’s important because the more information you get, the more the defaults need to be sensible. It does make you think that there needs to be some sort of protocol for it. So a device might ask “Who do I belong to now? What kind of alerts do they like to get?”

Adrian: Or they’d learn along with you. And it’d take you a bit of time to get used to each other. “Is this often enough? Do you want me to tell you more?” So every time you’d bring a new device into your life, it would have a settling in period where it learnt things. I suppose the ideal would be that the other devices would talk to it, and tell it to quieten down, almost.

Which is very human. If you came into a new office, and were being annoying, it wouldn’t necessarily be the Boss who tells you, because other people would tell you “Best not do that, ‘cos he’ll get angry.” You pick up the culture of the office. And the question here is, how do you pick up “the Culture” of your house, for a new device that has just turned up?

Hakim: This is very SciFi almost now! So I guess we’re talking about incorporating all of these devices. And the next topic on our Table of Contents is “Whole System Design” which is almost exactly that. And our subheadings here are “Online components” and “Playing to the strengths of each device”, which is about these devices playing off each other.

Adrian: My initial thought is that you use devices for what they’re good at. So Bubblino is very good at blowing bubbles. But no good at letting you type things into him, or giving you detailed information. Because he hasn’t got a keyboard or a screen. His entire output possibilities are: he has a motor, to turn bubbles. And he has 2 LEDs on the back that are part of the ethernet port. He could do with a bit more info to talk about what’s going wrong. Because if you can’t talk to the network, then it’s really hard to find out what’s going wrong. Is it that he can’t talk to the internet at all? Is it that he can’t talk to Twitter? Is it that he can talk to Twitter but Twitter is telling him to sod off. There are a few sets of things that could be wrong. And at the moment there’s no way programmed into him to find those things out. So at the moment I end up, if there’s a problem, looking at the lights of the internet stuff, and seing how fast they’re flashing. If both lights go off for a bit and back on, I can tell that he’s not getting an IP address, for example. But that’s not a good UI… because you have to be me and know how the code works, and how problems might manifest themselves in a couple of lights on an ethernet port, which is fairly detailed information.

Hakim: One of my favourite devices at the moment, and I guess one question is “is it a Thing” in the IoT sense, because it’s really more a “Thing of Internet” than the other way around, is the Huawei mifi device for the Three network. It’s a fantastic device. It has a very nice, minimal screen, and a button. And all the button does it switch it on or off. And all the rest of the UI is that it’s a router, and it has a web-page available at 3.home, and you configure it there. And it’s so much nicer than a suboptimal device with crappy LEDs and confusing buttons…

Adrian: Exactly: your laptop is very good at having a screen and a keyboard, and is so much better than a tiny device with a small screen and no keybaord at displaying and interacting with compelx information. That’s the kind of “playing to the strengths of devices”. So for example, you configure Bubblino by going to the Bubblino website. There are issues that, in some failure states, he can’t actually connect to that to communicate to it…

Hakim: And so he doesn’t have a webserver built into him? Is that something for Bubblino 2.0?

Adrian: I’m not sure if that’s a particularly friendly way to do it. I could have put all that configuration stuff on him, rather than on the web. But then people would have to be able to work out what his address is to talk to him…

People can just about go to 192.168.0.1, or 3.home or whatever, but as you get more devices, it has to be very easy to connect to them. And actually, my car stereo does a very good job of that. It’s an Empeg unit, a small ARM linux computer with all my MP3s on it. And when it’s not in my car, I can plug it into the ethernet at home, and I can run a small program, on Windows admittedly, to drag and drop files from my computer. And I presume it sends some sort of broadcast to my network to ask if there’s any Empegs around, and as long as they’re networked, they show up as icons to download tracks from.

So if we had some program like “find the Bubblino”. There are protocols like UPnP and Zeroconf and Bonjour to let you do that.

Hakim: It reminds me of something that Paul Kinlan, our resident Google Developer Advocate has talked about with WebIntents. You might say “I want Bubblino to be the thing that notifies me about tweets.” Or something else. And then let the network work it out. So, even… Bubblino might not need to know anything about Twitter, and some other device might do that.

Adrian: You could… though this leads in to one of my pet hates with the proliferation of middleware. Everybody and his dog wants to try to own the middleware bit. And I think this is a short-term win. There are cases where it’s useful. For example Pachube is useful to connect data to. So, depending on what the device is, you can connect things to it, and there are uses for it. But so many different services claim to be the webservice to rule the whole internet of things. And I think that’s just misguided, and not going to succeed in long term. For 2 reasons:

One: it’s something in the way, a new dependency in the chain. So as well as Twitter being available to check. And ok, Bubblino has to be a bit more intelligent to connect to tiwtter. And there are issues there, like Twitter changing to OAuth, and noone’s worked out how to do that on Arduino, and in fact it might not be something the Arduino is practically capable of..

But… if there’s an additional thing in the middle is down, or if the company that is building your Bubblino… so at the moment if you bought your Bubblino and then mcqn.com went away then you wouldn’t be able to update your Bubblino because the config is there, and can’t really be moved, but it would still be able to talk to Twitter… and sure, if Twitter changed their API, then you’d be kinda screwed. He still functions though.

But there are a whole load of Nabaztag rabbits that became “interesting paperweights” when Nabaztag went bust.

Hakim: I guess I’m really talking more about the “middleware” provider as having a protocol to allow devices to decide amongst themselves who does what. And the owner of the devices would have the final say in configuring that. So it wouldn’t be at the mercy of “some service”.

Adrian: I guess I’d have 2 requirements on that. It needs to be an Open Protocol. Which suddenly makes it harder, because you’ve got to publish it, and other people have to agree that it’d be useful, and you’e got to incorporate their feedback. And it’s almost the opposite of the Apple model where it kinda works nicely because Apple are in charge of all the hardware and software, which is Good, but it doesn’t work if you want to connect some other device that they haven’t thought of. And I can’t remember the second thing…

So I think the WebIntents thing would be useful. And it depends on how much intelligence you want to push into Bubblino, and how much elsewhere. Because he is just a sort of notification device that blows bubbles, so you could hook him up into a “notify” intent. So you could have a sensor that realised that there was a fire, and you’d get bubbles when your smoke alarm went off, instead of an annoying high-pitched noise…

Again, I’m not sure that’s the best reward/response to ”I’ve set fire to something and I might burn the house down… yay! bubbles!” It might encourage you to set the house on fire.

Hakim: So perhaps the “intent” would be split into “nice notifcation” and “nasty notification”?

Adrian: Yeah, so we could make sense of it. But IoT devices need to be first class devices on the web. So if Bubblino is a thing, and the only thing he does is talk to Twitter, then he should talk to Twitter. But not to some middleware service that makes it “easier to talk to Twitter” because he can’t really cope with it.

Because I lived through all the fun of that in the mobile world in the late 90s with WAP, because everyone decided that the internet was too complicated for mobile devices, so we need this thing called WAP, which is the internet dumbed down. And nobody ever used it. And 5 years later, all the phones were capable of talking to the real internet. And admittedly, even at the start, some phones were capable of that, because, well, we had a phone that was capable of doing the internet. And then we added WAP to it, because they’d persuaded everyone that phones weren’t powerful enough… so that’s a personal bugbear of mine.

Hakim: But we do still make things nicer for phones, with mobiles stylesheets and so on.

Adrian: Yeah, but partly that’s you’re just making the web more optimized. But if you go to a website that isn’t optimized, you can still visit it, and it might be a bit clunky, but you don’t get to a point that someone’s tried to turn your site into something “easier” for your phone, and it’s lost some of the more complicated bits that happened to include the really important thing that you needed. And sure, that bit might have taken ages to download, because it’s massive, but if you really need it, you don’t care.

As soon as you have a little ghetto of the internet, then you’re a little ghetto of the internet. And the internet won’t update itself for you. But if you talk to the source, then you get that info. And devices will get more powerful, with their equivalent of Moore’s law, the long tail of that.

So, right now, the Arduino is a little 8-bit thing, and quite limited, and it’s getting to the point that an ARM is almost the same price and has loads more memory and processing power. So maybe in a few years people will be saying “Oh the old 8-bit arduinos were cool, remember them? The ones we used to use back in the day?” and then everyone’s on the new ARM Arduino. Like the next Arduino Due will be. And then you’ll have enough memory and processing to be able to do https, so your arduino will be able to talk to Twitter. And suddenly anyone that’s doing middleware that “made it easier to talk to Twitter” because that was an interesting or difficult problem to solve will suddenly realise that it wasn’t.

So.. that was whole system design.

Hakim: yeah, so the next heading we have is “PC for configuration”, which I guess is really part of “whole system design”. And then “IoT device supplies the main interaction” which is really part of the same thing.

So the next question is really “Why Things?” and I guess it might be useful to kind of ask “What are things?”

So I was trying to tell my dad about the book, and then later I overheard him proudly telling a friend of ours “Hakim’s writing a book about ‘Things’” which was quite cute.

Adrian: Heh, so I tell people “Oh, I help build the Internet of Things” and they reply “Oh you build “Internet Things”… do you mean, like, websites and stuff? Things on the Internet?” It’s a crap name, but unfortunately we’re lumbered with it.

Hakim: I quite like the name actually, it has a magic quality, and fun sound to it. More than “ubicomp”

Adrian: I quite like ubicomp…

Hakim: That’s fine, but IoT sounds more…

Adrian: I guess ubicomp is a bit industrial, 70s. There is the “Thingternet” that someone was trying to popularize.

{discussion about domain names, and that so many of the good ones are taken}

Adrian: So we know that we’re in a field that is interesting, because we know people are domain-squatting. It does kind of validate it. We should have a service called “isthisstuffpopular.com” and it does a search, and if people are squatting on domain names related to it, then you know “Yes, we’ve arrived. This is something people think might be interesting!” An alternative to the Gartner Hype cycle.

Hakim: You could track that against how many are actually being used for real sites too…

Adrian: So, what are Things? It’s tricky because well, everything is a thing.

Hakim: Even computers and phones…

Adrian: Mike Kuniavsky defines it as computing power becoming cheap enough that you can embed it. So he talks about a 486 as being the first computer that was powerful enough to connect to the internet. And you can get that processing power for 10c. It was 50c in 2007 when I stole his slide to put into my slide decks to explain this stuff… and when you get to that kind of price, you can think of it as something you can just do if it’s convenient. And mobile phones are getting that way too. Yes, the mobile phone that you buy is a few hundred quid, because it’s got a couple of 1Ghz processor, and massive hi-res screen and etc. etc., but if you just want the 3G dongle of a phone, then those modules are £20 from Farnell in quantities of 1… so £20 to put a mobile phone into something isn’t a massive cost. And that’s dropping all the time.

So computing power is just a bit ahead of that. One thing is that it’s not always stuff that’s connected to the internet. Cos it can just be things that are a bit “smart”, they could maybe be products that have some sort of awareness. So… is the Glade air freshener that goes off when you walk past it an IoT device? And it almost is: it’s not connected to the internet… but it’s got the computing power to detect whether someone is passing, so that it can cover them in loads of chemicals…

And that processing power is cheap enough to put into a commodity air freshener. And compare that to bathrooms in pubs that are just on a timer circuit. But compare that to… urinal cakes. Which are totally dumb objects, where we know that the chemicals over them will tend to break down over a week, or whatever, when water goes on them. But now it’s cheap enough that we can put some electronics to measure things, and deliver information. So you start to think of computing power almost as a material, which is how Mike Kuniavsky defines it. So when you think about designing a product, you think about whether you’re going to design it out of wood? or metal? And if wood, MDF? or veneer? etc. And those are decisions you make about design. A nice veneer might cost 50p more, but it might look nicer, so you can sell it for more. And now you think, how much more can we sell it for if we put some computing power in it? The cost isn’t that much more.

20 years ago, you’d never have thought of putting a computer in something. 1, it would be massive, and 2, it would add £2000 to the price, so we’d have to sell it for so much more. But at a cheaper price… you can start to talk about whether it’s useful for your table, say, to be connected to an online service.

Hakim: It’s also not just about the price of the computer, but the total cost of ownership. so in the past, if you had a computer, which cost several million to build, and required a building to store it in, and several dozen technicians to maintain it, and swap tapes out, and program it, and move things around… Whereas today, you can put a chip inside a phone, and carry it around in your pocket or your bag, and drop it on the floor, and it usually won’t break or require servicing. So it’s also the total cost of ownership that makes it practical to be put in a Thing.

Adrian: Yeah, stuff lasts long enough that we can just throw it away and replace it with someting else… though perhaps that’s something else for our Ethics chapter… disposability and so on. Because mobile phones don’t have to last that long for most people who change their phones every 2 years or less.

Though perhaps we’re both atypical on that front… (laughs)

{gossip about our old battered N95 8GBs}

Hakim: So that’s “What” are things. But the reason we had the heading “Why things?” was … so you have a phone, and it has a keypad, and internet connection, and a camera, and all the input and output devices that you need, perhaps… so why would you want a Thing to be doing only a small part of that?

Adrian: I think the Manifesto for Furniture did a really good job of talking about this sort of stuff. Because yes, there is “an app for that” on new phones that have that App culture. But then you end up with a billion apps on your phone. And they’re all just tiny icons, laid out on a grid. And you can curate them and rearrange that and so on. But then you get to this point where you have to manage the whole assortment of functions that an app could provide, except you never get around to running it.

Whereas what Things are good at is: being in a set context. So, for example Russell Davies’s bike map that he built for the Homesense project is a really good example of that. It’s a little map of the area around his flat, and little LEDs for where the boris bike stations are. And if there are more than 5 bikes at a station, then the LED lights up. So as he’s leaving his flat, he’ll know which way he has to turn when he gets out, if he wants a bike.

And sure, you could do that with a phone, and it could tell where you were with GPS and ping the same API. But you’d have to pull your phone out and remember to use it as you’re leaving the house. And of course if he’s out and about, then that’s what he’d do. But when leaving the house, then that sort of ambient awareness is great. You can just glance at it, and get all the information you need.

Hakim: ok… so if you’re thinking about context, which is interesting. But if you think about Siri… which people seem to be much less excited about now that they’ve played with it for a few months, but it was supposed to have context like “Remind me when I pass the next supermarket that I need to buy milk” for example. So if you told your phone “Tell me when I’m leaving the house which way the nearest boris bikes are”, wouldn’t that be the same thing?

Adrian: but the problem is… how do you overlay that info? Sure, on that one scenario, then you might understand that beep… and maybe it annoys you every time you walk near the front door… unless it’s cleverer by talking to your door… (but then the door becomes the IoT thing) but if your phone’s going to do it on its own then it’d have to work it out by itself. So we could explore that sort of thing in the mobile world. And GPS directions would be quite cute for that. If your phone just vibrated if you were going off-piste, then you wouldn’t have to be constantly looking at your phone, but you could just get a warning when you veered too much away from the optimal path. So it could have some level of fuzziness about it… which is a nice way of getting away from screens.

Hakim: So, one of the things John McKerrell at DoES has done, which is his “joke” app, almost, is the Basic Satnav iPhone app, which just tells you if you’re getting warmer or colder, as you get nearer to or further away from your destination. And in a way you can imagine a lodestone device that physically gets warmer as you get nearer your destination. And wearable computing could be quite interesting with that…

Adrian: But if it comes down to context - your phone can beep, or vibrate, and that’s it. So it either beeps, and then you have to look at it to see why it’s beeping. It could be a bike, or a reminder. or a text message… and once you start to overlay all those things, then it starts to become a problem. When you have 1 app, or notification, or 3 of them, that’s fine. But what if you have 300 of them? 300 differnet tunes? How do you distinguish them? They could get more intrusive.

That’s one of the reasons for “why things?”

But it’s not an either/or. You might want a thing at your front door, or you might want the general phone device. Because you might not always be in the set context of an app.

{discussion about transport hackday + AIMES IoT project}

Hakim: so the next thing in our Table of Contents is “What’s your big idea?” which sounds really interesting, but I’m not sure I quite understood what it means.

The subheadings are “Just for you or for other people” and “Mass personalization / mass customization”. Do those work as subheadings?

Adrian: What I take “What’s your Big Idea?” to be is almost, but not quite a Call to Arms. Why are you reading this book? Do you have something that you want to build. So it’s something to help signpost what you should read through the rest of the book. Depending on the device that you want to build, you might just have an itch to scratch like “When is the next bus that leaves from near the office?” but you don’t care about building the device for everyone. Or maybe you think that a device could make your fortune, and you want to know how to build them for everyone.

So on one end of the spectrum, there’s just scratching your own itch. Building something that would be useful to you. Something that won’t take too long, that’s useful to you. So something for your motorized desk, Hakim, that prevents you from staying late at the office, by moving up and down at the end of the day to annoy you.

And at the end of the day, mass producing stuff… you’d approach things really differently, because building one thing for you is much easier than even building 2 things for 2 people.

Say if you’re building a website, you often fall into the trap of thinking you could spin it into a product or service that other people could use. And you think there shouldn’t be a big distance between “something you can use” and “something other people can use” but there’s actually a huge job of “scaling up”.

So some of it is just thinking about this sort of stuff. So, “If you’re just building somethign for you” then you might want to just read these chapters… but you might not need to read the ones about building PCBs, because your device might just live on a breadboard and that’d be enough. And you might need to read the web API chapter, because that’s important to be able to cobble up enough of a script to send an email when something happens. But building a scalable web service with a fully general API is something else. For just one person, you can put up with some hacks.

I’m not sure how much we’d signpost these things, but we could ask questions to think about while people are reading the book.

So there are alternate paths through the book… and alternate paths to building IoT devices.

Hakim: Yes, though, as we discussed before, perhaps we don’t really cater to the path of doing stuff for efficient, corporate, industrial processes? Because there are other resources for that.

Adrian: Yes. Though some things will be blurred. Because for example, you could imagine a task like booking meeting rooms, and make it more efficient, but using a really nice device on your meeting room door. So it’s partly about efficiency, but partly about being fun and beautiful, and for humans. So it’s useful to think about these things.