[sashahilton]

All these new standards looking to become the successor to WiFi/Zigbee IoT devices, yet every single one makes the same mistake - they think that because they find ways to force certification, take away end user control and extract licensing fees, that they’re somehow going to convince people to buy it as the next big thing.

I was cautiously optimistic when Matter/Thread was in its early days, but predictably as with most of these industry backed standards it’s turned into another pay to play walled garden. The CSA seems to be particularly bad at this.

Cant wait to see in subsequent years all the additional e-waste when manufacturers consider devices obsolete, and no one can repurpose them because of Matter spec mandating secure boot.

[andsoitis]

To succeed, consumers need to be able to rely on standards conformance.

Certification is an obvious way and that costs money.

Is there a way without certification that results in high conformance that consumers can rely on?

[sashahilton]

The standards side of things is true, however this can be largely solved by providing a reference implementation, given that no device manufacturer is going to implement the stack from scratch. Automated testing of firmware would also work. As for high conformance... WiFi and BT devices manage to work well enough by simply buying a tested chipset and building on that, no external testing/fees necessary.

I understand the certification if a manufacturer wants to sell a product commercially as 'Matter Certified'. For hobbyists or smaller players, pulling the reference implementation, loading it onto a cheap MCU, and calling it 'Works with Matter' would suffice.

As it stands, the latter isn't an option, because of the codesigning they've shoehorned into the spec. And for all the noise made about security, once connected to the hub the manufacturer can run whatever they like on it and send data back to their servers with very little visibility to the user.

Thread is arguably the interesting part for low power devices, and doesn't force certification. Matter is little more than a protocol spec, at the tradeoff of locked down devices and annual fees. For Matter over WiFi, I can't see any point whatsoever in using it. And for the costs of Doing Matter/Thread certification most smaller hardware startups will balk at the hundreds of thousands required to do so, and stick with WiFi/BT/Zigbee/Thread + roll their own protocol/app.

[doug_durham]

Having been in the business of creating standards compliant equipment in the past, the problem is not as simple as you state. You can provide all of the reference implementations you want and you still will get variances.

The state-of-the-art solution is to put a bunch of people on planes and burn a bunch of jet fuel to attend a "test fest". You can't issue interoperability until you do this. This costs money that needs to be paid by someone.

[ur-whale]

> ou can provide all of the reference implementations you want and you still will get variances.

A proper reference implementation should first and foremost come with an extensive battery of regression tests, something many a vaunted "standard" utterly fails to provide, being instead tomes upon tomes of impossible to decipher specifications, written - of all things - in human language.

Such a battery of regression tests, if properly design ought to take care of your "variations" in fairly short order.

[doug_durham]

In practice they never do. There are too many corner cases that people don't think of. It's almost a law of nature that you can't capture the complexity of any non-trivial specification in regression tests.

[EvanAnderson]

Certification is desirable to me, so long as it doesn't mandate taking away the freedom of the owners of devices for local control and the ability to install whatever software they want on those devices.

I had no idea that Matter/Thread mandates secure boot. Presumably that's secure boot without end user freedom to load their own keys. That's no good.

[rpcope1]

This is the Z-Wave approach and is honestly probably why IMO it sucks the least.

[Nextgrid]

Standards conformance is one thing. But to guard against manufacturer end-of-support and/or bankruptcy there must be provisions for adversarial interoperability too, which I believe is currently lacking. This is not the case in other, even commercial systems like KNX, where while the configuration tools are paid and proprietary, communication between devices works based on well-defined types and “registers” and any device can interact with any other regardless of the manufacturer’s wish, ensuring cross-manufacturer compatibility even in case of manufacturer failure.

This is why all the (commercial!) high-end home automation uses KNX, because installers can mix and match products and know they won’t get cornered even if a single manufacturer goes out of business.

(And if you want a cheaper system that can be operated with fully-free tools, you’re already covered, that’s ZigBee! But even KNX is “open” once you buy the license for ETS - the config tool - and then you can go to town on your system and reconfigure it at will.)

[lucideer]

> Is there a way without certification that results in high conformance that consumers can rely on?

Sibling commenters have already answered this in the affirmative but a better question is: what examples can you give supporting your thesis that certification drives high conformance?

I'm likely in a bubble but most certification systems I've seen exist in highly "competitive" landscapes (read: competing standards) & exist purely to support an ecosystem of institutions relying solely on certification to make money & providing no other value back to the system (e.g. in the form of standards advocacy, conformance testing or even standards maintenance). To the point I've come to anecdotally equate "certification" with wild-west/scammy corporate practices.

So as far as I've seen certification often has the opposite effect that you propose. At least in the IT & software space.

[naasking]

> Is there a way without certification that results in high conformance that consumers can rely on?

Open source a reference implementation and a conformance test suite. Open, transparent, and low cost. Also, don't lock out devices that aren't blessed.

[intrasight]

Why not have it be like the law profession? 100% open source with licensing.

[surajrmal]

The model seems to work. Is there precedent for the alternative working out well?

[EvanAnderson]

Working well for device manufacturers who want users to forklift out gear every few years?

I have been avoiding "IoT" in my home because I want stable 20+ year lifetimes for protocols and standards. I want to know that the outlets I hard-wire today will be controllable with whatever software I choose in 5, 10, 15 years. I want my thermostat to continue to have all its "smart" features for the lifetime of my HVAC system. I don't want separate "apps" for my washer, dishwasher, automatic water shut off, etc. I don't want Internet connectivity to servers that may be turned off at a manufacturer's whim to gatekeep features (or worse, basic functionality). The market is dysfunctional.

[simonjgreen]

You’re in luck! That’s Zigbee.

(Not IoT)

[hamdingers]

Does a device that provides an interface via a HTTP server need to pass any kind of TCP or HTTP certification?

Maybe I'm making some category error here but I can think of dozens of protocols that do not require certification.

[thequux]

As much as I dislike mandatory certification, I can understand the need for it in wireless battery powered devices: a malfunctioning decide can talk the battery life if everything within range, and most consumers aren't equipped to realize that this is happening much less identify the device that's causing the problem

Perhaps the solution is to make the spec open but make using the trademark contingent on certification (much like USB, for example)

[ur-whale]

> The model seems to work.

It "works" in the sense that it excels at separating implementors from their money and locking up end users in a cage, sure.

[ajross]

FWIW, the "Suzi" thing they're talking about seems like an attempt to compete with LoRa, which while technically an open standard is entirely dominated by Semtech.

It's not clear to me why the manufacturers aren't just making LoRa radios. This feels like an xkcd 927 situation to me.

[teleforce]

The sub-GHz Suzi is for LPWAN or Low Power Wide Area Network, similar to LoRa, SigFox and Wi-Fi HaLow.

LoRa PHY is proprietary by Semtech but LoRaWAN the data link layer is open, that's why the manufacturers aren't just making LoRa radios.

SigFox is full-stack proprietary LPWAN solution but the founding company already went south.

Currently the main open standard alternative to LoRa is IEEE Wi-Fi HaLow.

The main problem with these LPWAN standards, however, is that all of them struggling with limited Line-of-Sight (LoS) or Non-Line-of-Sight (NLoS) LoS environment that's very common in LPWAN. For example LoRa is very low power but its limited chirp spread spectrum (CSS) based modulation is bad for any limited LoS environment. In addition, its spreading factor (SF) only work for a single channel per user.

I'm now working on the LPWAN PHY alternative that's much better for very limited LoS than LoRa. Looking forward to benchmark against Suzi.