@Hugo, Think about it this way. Before IoT anything that had a connection to the internet via Wi-Fi, cable or cellular could do anything IoT does. From this standpoint IoT brought nothing new to the table. What It did bring is super small embedded devices with built in connectivity and low power like your IMP. If you go back just a couple years there is no way I could get anything the size of the IMP001 and be connected to the internet. I would be forced to use a much larger platform, using more power and probably wind up being the size of a pack of cigarettes.
Well, whilst it could in theory do anything the IoT does, it couldn’t really in practice; eg before my definition of IoT, you couldn’t just pull on a weather API to get weather, send data to a graphing service to quickly visualize it, send a text message with Twilio, or get your geolocation by sending BSSIDs to google.
The critical difference is that whilst almost any application was possible with communication (doesn’t even need to be IP based… plenty of M2M still uses SMS), nowadays you can knock stuff up in no time at all and pull on a huge range of resources that are built, ready, and largely plug & play (due to the standard nature of RESTful APIs). That’s what really makes IoT powerful - the integration with the internet, not the fact that it uses IP as a communications medium.
ie, my definition of IoT is around interconnectivity with other services; yours is more around hardware evolution.
IMO, obviously!
Been following this thread with interest. This has often been discussed when trying to describe point to point telemetry (p2p), machine to machine (m2m) and more recently internet of things (iot). In my view much of the differences and similarities between the three relate to the hardware and infrastructure (or the systems architecture) used and then how data is transferred from source to the end user. I also think the use of independent or external web service API’s to deliver a solution is becoming a key differentiator between something that is IOT based versus other types of messaging, data capture and data transfer systems / methods.
@hugo, Just know I am just having a friendly debate about this, I am not arguing.
For me personally, when you said:
Well, whilst it could in theory do anything the IoT does, it couldn't really in practice; eg before my definition of IoT, you couldn't just pull on a weather API to get weather, send data to a graphing service to quickly visualize it, send a text message with Twilio, or get your geolocation by sending BSSIDs to google.
I was doing things just like this more than 7 years ago with what was out there back then. Getting the weather, sending data to see it live, and so on. Here is a video I made 4 years ago that streamed audio from a SD card on my embedded device to the browser live. Did the same thing for temp sensors and so on. You can mimic being a server or a browser. It just take time and a packet sniffer.
Today people are making things much simpler doing the same thing I did way back then. This is just technology moving forward. We could do the same things today as we could before IoT. The only thing that has truly changed is the size and power of the device.
Yep, like I said the difference IMO is that you don’t have to do it all yourself any more. You put together building blocks.
But don’t you think that is a software thing and not a hardware one ?
I mean, would you consider my PC, laptop, phone, or tablet a IoT device?
They can all do the same thing as IoT using all this new software out there.
I would not hang a PC in my garage to control the garage door from my phone, I would slap a IMP there to do it. Small, compact, low power.
There’s very little new about the hardware; I could have built an imp (similar size, maybe 2x the cost) in 2006 using commodity chips. However, back then there was no twilio, no plotly, no firebase, no pubnub, no mobile phone apps (well, not what we’d call an app now), no keen.io and generally no data analytics-as-a-service.
It’s the ecosystem that enables IoT to be a real thing - otherwise it’s just a small computer. Like you I’ve been building small computers for many years. Even the mp3 player myself and other imp-loyees shipped in 2000 had ethernet on it 
Wow, you could have built an IMP001 back in 2006? You a better man than I. I don’t know of anything out there back in 2006’ish that would fit inside a SD card housing.
Let alone the Wi-Fi. I think we can both agree one one thing though, IoT is going to be the hottest thing.
Back in '06 we were using the Marvell 8686 in a CSP in the iPhone… there were also CSP microcontrollers, though 1MB of flash was likely a bit of a tall order on-die at that point
IoT will indeed be hot if it can be deployed… and platforms are really key to deployability across a wide range of industries. Hence, electric imp!
As @hugo rightly says it is all about platforms… and I note @seulater’s earlier comment that he could use a WIZnet Ethernet module to solve a problem.
So why not join the two and make an Imp based Ethernet module. I could certainly use one, as need the imp platform but wifi not always suitable in some cases.
Good. I look forward to seeing the Imp Ethernet device. Hope you go for it sooner rather than later as nothing much in that space at the moment… opportunity knocks.
I only know these chaps (https://nabto.com/) who provide a secure web server model through Ethernet connection to device.
Would be curious to know if there are other Ethernet devices out there which can be programmed via the cloud etc.
You likely don’t have very long to wait before there’s an interesting announcement 