Hacker News Reader

Comment by awalton | Hacker News Reader

Proving yet again that the "Internet of Things" won't work until the "Local Area Network of Things" works. If the "thing" requires the Internet to work, these kinds of failures will continue plaguing them for eternity.

Sadly the "IoT" market is practically predicated on selling some hideously expensive service that you don't want or need on top of a piece of hardware that would be perfectly happy working against a home edge gateway rather than "the cloud". Approaching a VC asking for money for digital lightbulbs and dimmers that work against a local server running on a Wifi-router-like-box? Using a standardized protocol, so you don't need 10 different home edge gateways to orchestrate different tasks? Get out, we don't even want to hear it...

Accessing the Internet should add features - not having the Internet should not burden or brick the device.

View on HN

Except consumers want the ability to control/check their IoT devices when they are not at home. Most homes in the US do not have a static IP address. Yes, you could set up OpenVPN on a router and use DDNS, but you would still be reliant on a 3rd-party DDNS provider that could always go out of business. Also most home routers supplied by broadband providers do not support OpenVPN.

An admittedly expensive solution would be to use the Ethereum main network to communicate with the IoT device.

A middle ground is have a local-first control, that the company exposes to the internet via the customer's account.

For example, a local administration page that can be accessed within the LAN, and which can then be proxied by the provider. In the case that the provider goes out of business you lose the remote control (unless setting up your own VPN) but the device won't be bricked.

Unfortunately, I imagine some companies intentionally avoid doing this to require customers to pay a subscription fee.

I still don’t think that really avoids the issue of devices losing functionality if the company folds. If an IoT company goes OOB, you also lose things like security updates, tech support etc. I would give an ‘orphaned’ IoT product a year tops before it was no longer safe to have on your network.

I think consumers need to understand that when they buy a connected device, they are also buying a license to the software that powers it. That license is generally not perpetual unless specified.

Most of these companies don't have any subscription fees. They just provide the "free" cloud service for as long as it takes to make a whole bunch of sales, then shut it down and suddenly all the devices stop working. if you're selling a $50 device that people have to wire into their home, with an expected lifetime in years or decades, it's pretty negligent to not provide for a sustainable service.

You’re saying it would be more responsible to charge a monthly fee? How could you compete against those firms that do not charge fees?

I'd prefer they just let you run the hub/central service yourself, with an option to have them run it for a fee, yes.

I think we'll start to see regular consumers harmed by some of these start-ups destroying hardware when they go out of business, and perhaps people will stop buying them unless there's some protection. Then the "free" solutions will be looked at suspiciously and hopefully discounted by the tech press and retailers. Alternately, retailers might be held liable for refunding purchases of products that stop working before the legally mandated warranty date (such as in the EU) or by credit card companies in the US, so they'll stop selling any cloud required garbage.

It’s possible but sounds pretty optimistic. More likely that the big 5 tech companies (who can pay for cloud services to function indefinitely) will consolidate the smaller IoT plays, and offer users a limited feature set for free, with premium tiers as well.

Also remember that if you’re maintaining the hub yourself it won’t get security updates.

Home Assistant on a raspberry pi gives you an open source home hub with a mobile app. It has built-in dyndns functionality from a variety of providers, using an open standard. It supports on-board ssl via letsencrypt, and provides easy instructions for forwarding the relevant ports on your router. I think the router forwarding step can be fixed with upnp? Then you've got a home hub that works from anywhere and doesn't require a specialized cloud discovery service.

I'd love to see someone provide a pre-packaged zwave/zigbee talking home assistant appliance with support for snips.ai voice control. Unfortunately, hardware-wise it would probably cost $250 to build out of off the shelf components, but perhaps at scale it could be affordable? Google and Amazon are selling their closed, cloud chained voice assistants for practically below cost, though, so I don't know how such a product could really compete.

I don’t foresee any company taking a product like this to scale. If an amateur can build it out of OTS components there is likely little there that is defensible, and you’re right the space is competitive.

Reply on news.ycombinator.com