Matter, set to fix smart home standards in 2023, stumbled in the real market
arstechnica.com
arstechnica.com
Companies don't really want to give us freedom of choice. They just want to take our money. Ideally through subscriptions so they have to do almost nothing for it but can cash in easily.
So all the "special sauce" doesn't have a technical reason but am economical one.
We at AirGradient see that very well that more and more companies force their users into the cloud and some monitors even refuse to show current air quality when not registered to the cloud. Imagine you would need to register your fever thermometer with an app in the cloud before you could measure your temperature...
Part of the problem is not only the investors in Matter but also VC funded smart home equipment manufacturers that are pressured to go down the recurring revenue business model.
At AirGradient [1] we want to demonstrate a counter example. Our fully certified air quality monitors are open source hardware and thus we provide the firmware code and they can be easily re-flashed with whatever software you like to use. With this openness a great community has emerged that maintains integrations to various systems like Home Assistant via ESPHome.
I don't have an immediate need for one of these, but do you have any plans to expand to Europe or work with a local distributor? Inflated shipping cost & time, VAT & customs duties (+additional processing time) and a cumbersome return process if anything goes wrong, makes for a less than ideal shopping experience.
At the moment we still have a lot of dynamics in the product, eg frequent firmware updates on the shipped units and this is why we currently prefer to ship directly.
Once things are more settled down we will look for fulfillment centers in Europe and the US.
Thanks to Achim for all your awesome work, by the way.
There isn't an HN promo code is there?
See: https://forum.airgradient.com/t/airgradient-integrations/703...
This way you’d get a list of potential customers and we’d get a notification when we’d get an easier option to buy your products.
We socially infantilized each other. Technology plays a huge role in that.
Voices of the past warned that extreme division of labor would result in humans that are lazier, stupider than the dumbest animal. People sit online screeching about the slightest social faux pas, do nothing about intentional destabilization of education, healthcare, and infrastructure by political parties.
We complain that every last mile problem, every edge case we encounter, was not solved already by Big Tech. Never occurs to us to solve them ourselves.
I grew up being taught how to rebuild cars from the frame up, fix houses for people in need struggling in Midwest winters, was mentored in physics, chemistry by college age aunts and uncles, had a grandfather who retired after decades of designing computers for IBM teach me analog circuit designs/implementations, and helped farmers birth/raise livestock, got a couple uni degrees a while back. I'm barely 40.
Meanwhile my office worker colleagues waste time and real resources driving for an oil change. They demand others solve their survival problems for them. Friends struggling in this economy have started chanting "If I'm just going to berated for not working 7 days/week for a rich person, may as well quit and work 7 days/week growing my own food, fixing my home."
Fingers crossed the fiat hallucination Boomers force us into serving is about to vanish with them. Physics does not put an obligation on the future to propagate their rhetoric.
I don’t drive, as it’s been largely both unnecessary and impractical with where I’ve lived in the past decade, but I know how to check and change the oil in a car, among other basic vehicle maintenance tasks. It’s not hard or time consuming.
When I owned cars I often used to fix issues myself (as I used to drive old ones I've replaced entire window mechanisms and wiring harnesses) but not the oil changes. It's just not worth it. And it's so much easier with a full hoist, I only had portable jack.
As for disposal: in a fair few places I’ve lived the city has had regular free pickups or drop off locations for “hazardous” domestic wastes such as oil, car batteries, etc.
IMO the real issue is lack of instilling a sense of agency at a young age.
You were taught how to build cars from a young age. I’m sure you were encouraged to try things yourself along the way. You learned that your actions have a direct impact on the world and that you have the ability to adjust your actions to adjust the impact.
(At least in the US) Education is so retention and information focused that we don’t teach _how_ to apply that information.
Because of this we grow up looking to others for how to do anything, even if we have the know how.
It’s usually not until later in life when people realize their actions can truly have an impact when applied correctly
You could have used a relative educated in economics. Remember that just because you’re an expert in one field doesn’t make you one in another.
> If Matter can succeed in convincing companies to compete on the merits of their devices—not the incidental revenue that comes from being a single-system device provider, or phone app marketing...
The sentence you (partially) quote starts with "if" and ends with "wish". Cough.
You didn't make a point yet.
You claimed I misunderstood, then quoted a sentence that starts with "if".
In similar vein ESPHome and Tasmota are 2 independent FOSS projects that can replace proprietary firmware on many commercial smart devices based on ESP32.
Some WiFi-based smart devices like those made by Shelly work without internet access and integrate with HA via local HTTP API.
Fully local smart home is possible, but you'll need to research what you're buying - as a general rule, you should avoid WiFi-based smart devices unless they offer a local API. Everything else (ZigBee, Z-Wave) should work offline without proprietary hubs, though I'm sure there are exceptions.
BK72xx: BK7231T, BK7231N RTL87xx: RTL8710BN, RTL8710BX
https://esphome.io/components/libretiny.html
But yeah, using both Tasmota and ESPHome I’ve moved all (bar one) of my “WiFi Only” devices to being fully local devices.
(The exception being my thermostat for my heating, that uses a tuya-mcu and a LCD driver IC so I don’t really want to reflash that and faff with the tuya-mcu this side of winter, and I’m planning to replace it with a OpenTherm controller next year anyway. In the meantime I’m using a local tuya integration in HA to just send it commands directly)
It's good to see that people are able to hack this stuff and make it less reliant on the companies. Heck they're probably selling some of it at a loss and hoping for recurring revenue from usage stats and other spying.
The bulbs aren't as good as Philips Hues though.
Convenience ? One minute of me debugging a http request to fix my light is evaluated dearly.
If that happen once it will offset years of voice light off/on and other lazy ass of poor design workaround.
Not just with HA, but tech in general really. I love the idea of FOSS, but it's important that stuff just works, every time, without hackery.
But I still wanted to use Home Assistant to control lights. The solution? Smart light _switches_ and wall plugs.
I went with Lutron Caseta wall switches and plug-in dimmers. (Expensive at $60/switch.) Lutron has been doing commercial automation for a very long time and it shows.
1. They work without any automation. Most dimmable LEDs will work.
2. The remotes connect directly to the switches and don’t need a hub.
3. The pro hub model connects to home assistant. You can use unpaired remotes as switches. (Using the correct blueprints.)
I’ve had this set up for over a year and I’ve yet to replace a battery on the many dozens of remotes I have. And I’ve never had a problem with the switch not working.
I get to do all the tinkering I want without breaking my house. :)
Note: I do have some Phillips Hue bulbs for color but those are in lamps separate from overhead lighting.
I think that is because the characters tend to have a very nonchalant way of using it: Whatever terminal is nearest can be used to interact with stuff in vicinity. There are no boundaries between devices, services, or the like; everything communicates with everything else, over any channel available, without further thinking.
If we ever get past showing off how the ceiling bulb can do red light, maybe we could start having a protocol for automatic device capability negotiation/discovery?
But after that, wouldn’t it be great if you didn’t have separate devices anymore, but a single digital identity across all of them, with an interface to control unique capabilities of each of them?
For a banal example - there’s no reason we shouldn’t be able to set a timer on our phone, and check it on a computer, or kitchen counter display. On the other hand, I can’t check on the progress of my vacuum robot from my laptop, because the vendor only offers a mobile app. I can’t pull the stats from my car at all, I can’t check a hard drive without plugging it into something compatible, I can’t easily send a big file to someone nearby. Technology should allow us to do all that.
Devices should promote their capabilities and offer a standard interface to control them, and simultaneously be able to display controls for the interface of other devices, depending on available means of input. Done well, this can actually be zero config.
Open source could do it in theory, but only for nerds. Open source can’t create mass adoptable products because making technology easy to use is fabulously expensive and difficult.
I don’t think the market alone would create a unified rail or telephone network. Yet that has brought immense benefits for society at large, and vendors still manage to capitalise on top of the plumbing.
if you have an account online somewhere you can log in from any device. the device itself is merely a machine to run the code downloaded from your online account. you could do everything though that account and treat the device like a dumb terminal.
any specific device that provides special functions, like your car, could also connect to that account and upload its data there, so you can access it from anywhere else.
technically this is already possible today.
I wouldn't mind connected services, but when my data is held hostage and the one in control is a faceless company, it does not feel like technology anymore, something that would help you accomplish your task better. Take my smart tv, I would be glad to connect it to my network and not bother with the apple tv when I want to play a movie, but the last time I tried, it tried to forcefully upgrade itself and wouldn't let me skip some onboarding screen that tried to get me to create an account. I had to unplug it. Or when some headphones or mouse manufacturers want me to use bloated proprietary apps just to configure them.
While what you describe is technically feasible, it never feel like this as tech companies always start a conflict with their (paying) users by taking their data hostage, enshitifying their platforms, and destroying any attempt of integration.
i don't know if my idea how to do this is to simple, but essentially, i think what is needed is the ability to connect all my devices into a virtual network, where some of these devices act as connection portals to allow me to interact with them from any other device. i think the hardest part is the security of these remote devices so they don't leak my credentials to others using the same device later.
And this should be across domains. A data or diagnostics scanner for example should be able to retrieve data from my car, from my fridge and from my hear rate monitor.
I understand how compelling the marketing hype seems at first, but a quick investigation reveals this to be yet another mafia shakedown and effort to lock down and prevent any meaningful competition.
They have a 16-bit Vendor ID. That's 65,535 vendors supported. Globally. Ever. One must pay the CSA ([0] $2,500 per PRODUCT or $7,000 per YEAR) for one of these artificially scarce identifiers.
In order to download specs, you're expected to accept TOS, authenticate and then you're emailed a (presumably authenticated) link to the PDF [1].
You need to wait for the consortium to define every single application layer detail for each and every type of thing [2].
Of course, the decision makers are unable to make high quality decisions since their own businesses depend on trapping customers in their respective walled gardens. The goal is to prevent any small or medium sized businesses from competing with the small handful of multinational megacorps.
From what I can tell, it's all of the worst parts of USB, without HID and Serial to fall back on for things they didn't think to define.
[0] https://csa-iot.org/become-member/#Membership
[1] https://csa-iot.org/developer-resource/specifications-downlo...
[2] You must download the "Device Library" spec yourself, as my link is fingerprinted and I do not wish to violate a hidden TOS clause.
[edit] formatting
I didn't know any of this. I pay attention to IoT (and run a modest Home Assistant setup in my home.) Had I known about these constraints, I would have written off Matter from the beginning.
What were they thinking?
As soon as you need to instantiate something with atoms, the cycle-times and overall difficulty increases exponentially. Just when you think you've got something, all these despicable gatekeepers shove their hands in your pockets.
There are lots of reasons we can't have small batch, niche electronic products, and these license fees combined with even worse regulatory costs require enormous sales volumes to amortize.
It's really concerning to see the same rent-seeking values making such strong headway in the software space over the past 5-10 years too.
Take a gue$$
This is why I love HN, thanks for pointing it out.
Isn't that the same as USB or PCI? Both also have a 16-bit vendor ID.
The USB-IF also artificially limits the vendor address space, which gives them the ability to charge exorbitant fees before anyone can deploy a device without conflicts.
I haven't tried to build and deploy a PCI device yet, but given your point, I assume the same game is being played.
At least with these 2 technologies, they can lean on being around for some decades. Bits were quite a lot more expensive 20+ years ago.
Another measly byte (or two) could have made it possible for fees to be reduced down the road, but as-is, there is an artificial barrier to popular access.
WiFi has gotten way more reliable, and devices with open local APIs are more common now. Bluetooth is even starting to be used occasionally. YoLink's LoRa based hardware is still better than anything else even slightly affordable, even though it's still totally a closed cloud based platform.
I assumed my link in my email was fingerprinted as there are some odd characters encoded into the URL.
As it turns out, it matched the search result link.
That's good to know.
Thanks!
Zigbee is also pretty cool, between these and an odd Wi-Fi only device here and there, it seems more than adequate.
The protocol is also excellent for battery powered devices in terms of super low power consumption: depending on the device's needs, the protocol allows for the device radio to be completely off for hours at a time. For example, I have a couple of key fob controllers that we use daily but we haven't replaced the batteries in them for years.
My only fear is that "consumerist" IoT platforms that change every few years will usurp it and new Z-Wave devices will become unavailable.
I hit a problem where it was really difficult to debug setup / configuration problems.
I suspect this was due to zwave devices being as cheap as possible: my devices had just 1-2 buttons and 1-2 blinky lights for configuration.
I would have gladly spent $30 more per device to get better device diagnostics.
It really just seems to come down to your local market. Zigbee seemed to have more compatible devices in the US vs zwave so I went with zigbee. I only have about fifteen devices, mixed between a few manufacturers, but it's been rock solid after I switched from my initial conbee to a sonoff dongle.
All wired Z-Wave devices also act as repeaters to extend the network, which is handy.
The Z-wave devices turned out to be significantly more unreliable surprisingly.
The only z-wave device I still have running is my smart lock. Everything else has been replaced with zigbee.
Also important: not overdoing it. I don’t have tons of time to tinker with devices, HA, etc.
For such lights, the permanent power usage of the controller can be much larger than that of the light, even if you accidentally leave it on way longer than needed now and then.
The EU limit allows each controller to use 0,5w. See https://commission.europa.eu/energy-climate-change-environme.... And they do use close to that. The Philips Hue bulb uses 0,4W when off. See https://www.anandtech.com/show/6805/philips-hue-automated-ho....
So, that’s about 10Wh a day. If you use that to control a 10W light, you’d have to decrease its ‘on’ time by a hour a day to come out positive in energy usage. I have many lights that aren’t even on that long each day.
I think home automation of lighting fixtures as a way to conserve energy made way more sense before LEDs became common.
You can't choose which device on the network is the controller when setting up a new device to a network.
For instance if you have a home assistant blue it might choose your Amazon echo to initiate the network. Therefore you have multiple networks running which may not respect your privacy or choice.
On the other hand some manufacturers are now actively fighting home assistant and the matter switch is non existent. A total zero number of devices in my home run over matter, something that I didn’t expect given the huge amount of excitement early on.
It feels like rather than everything slowly converging I now run two hubs still, custom smart home apps, a largely unreliable and useless home assistant next to some things with HomeKit.
I really can’t stand it and I’m incredibly frustrated by it.
I just started using it and getting into home automation, and that’s disappointing to hear
My two cents: thread and matter has the potential to be more reliable than WiFi. The biggest blocker is that thread border routers have no vendor interoperability (you have to buy a bridge for every vendor)
Why? They all work ~fine on zigbee which is pretty much the exact same thing just with even less oversight.
I haven't bought into matter/thread because I haven't found a single device I want yet.
And with that in mind IPv6 feels like a decent choice.
If vendors don't respect that foundation then the whole point of Matter+Thread is just gone.
[0]: https://www.reddit.com/r/Hue/comments/cpbu9w/ikea_tradfri_bu...
There certainly are issues with the zigbee devices, but that is not it.
I’m a firm believer in matter and threads future, but ipv6s home network mess is going to hamper it for years imho.
ICMPv6 (type 133) Router Solicitation (RS)
ICMPv6 (type 134) Routing Advertisement (RA)
Which my xfinity provided router does not. I spent weeks with support emails between xfinity and eve trying to get it sorted out before eve declared it an unsupported router and refunded me the purchase.
But for 90% of the population home automation is as elusive and as prohibitive as quality home espresso - expensive, delicate, and requires all the attention of a niche hobby. Not that most people aren't smart enough to figure it out, but for most people the juice isn’t with the squeeze. All the COTS hardware for easy implementation is a closed cul-de-sac and rather mediocre, and is condemned to be a neat gimmick that quickly loses its luster.
It's possible you've been scammed. I suspect there's a big market in selling knock off Hue bulbs because they can mark up the price so much more.
I guess a loophole might be to make the border router matter compatible but none of the actual devices. At which point you really shouldn't be buying them anyway.
Matter is an interoperability standard so that devices expose a common API. It works over WiFi, Bluetooth or Thread currently.
Thread is a network protocol like WiFi or Bluetooth.
The QR codes are used for pairing devices to a Matter controller.
The controller could talk to the border router if needed but that should be it.
Zigbee relies solely on noone being able to MITM the pairing process, where the security lies in the short window of opportunity and short range.
But Matter (and thus the Thread devices we talk about here) has an out of band mechanism for sending keys, since it requires QR + BLE.
I guess your point is that you typically can't form a Thread mesh between different border routers of different vendors. Which is unfortunate, but solvable. Don't buy/use border routers from such vendors. Because that problem is not the same as not being able to use a device from vendor X on a non-X border router.
Big advantage of Matter/Thread is that don't need a hub with its software. It is IP-based protocol. On Wifi or Bluetooth, your phone talks directly to the devices. On Thread, you need a gateway to translate from Wifi to Thread. Your phone only needs to know about Matter.
I don't think Matter devices talk over the Internet, I think they use non-routable IPv6 addresses. You can use HA to control over Internet but that uses the HA app. I don't know what other apps do, probably the same custom thing as with Zigbee.
Matter works similarly to HomeKit, with direct access to devices, but Matter adds the low-power Thread. It is also standardized so that could have generic Matter app instead of every device being its own app. And can add support to HA instead of custom Wifi and Bluetooth devices.
Some Google Home devices can be Thread hub, and Google Home can do Matter directly. iOS 16 has Matter support but not sure what that means. That may be the advantage that don't need to download app for each device but will be builtin to OS.