[bell-cot]

I'd say a temperature alarm - if I happen to feel I need one - should be a physically separate little thing. With it's own programmable temperature range. Usually it's in the fridge. But during a cold spell, I can adjust and move it to a spot where I'm worried the pipes might freeze. Or turn it into a "is the A/C working?" watchdog during a heat wave. Or, I can buy several to cover all those. But they're all one make & model, so I don't have to install/learn/use 5 different crappy apps for 'em.

And that way the the fridge can't die...and quietly take its built-in alarm with it.

[darreninthenet]

I said it was handy not something I would go out of my way to get... the fridge has it so it's useful. I don't understand why so many people on here are so offended by a useful feature that from most people's perspective is just a feature the fridge has rather than one they shopped for.

[arter4]

I personally am not offended by a feature. I wouldn't march on the streets to denounce this or anything like that.

What I am saying (though I cannot speak for anyone) is that I, on the contrary, would go out of my way to not get, not because I have thin foil hat or whatever, but because the reliability and privacy concerns this kind of thing carries far outweigh (in my scale) the benefit of having something that will come in useful once in a decade.

Note that things like that have already happened.

When an AWS region went down, things like doorbells and lights stopped working properly (https://financialpost.com/technology/personal-tech/pitfalls-...). Which, if you think about it, is ridiculous. I cannot turn my lights on because somewhere, potentially thousands of miles away, some datacenters failed, even if those datacenters could, ostensibly, only give me ads and "improvements" and not any core feature.

And when it comes to privacy, Roomba employees posted on Facebook picture of a woman on the toilet, a picture that was taken by the Roomba itself while cleaning (https://www.technologyreview.com/2022/12/19/1065306/roomba-i...). Now, you might say that the Roomba algorithm has to train a lot to be able to work properly, but why take pictures? And why upload them to the Internet?

Of course everyone makes their own judgement calls. I'm not offended if anyone gets smart fridges or voice activated lights.

[BoxFour]

> I cannot turn my lights on because somewhere, potentially thousands of miles away, some datacenters failed, even if those datacenters could, ostensibly, only give me ads and "improvements" and not any core feature.

It appears there's a misunderstanding here. When smart lightbulbs lose internet connection, they still function as regular lightbulbs.

In such cases, you simply revert to the traditional method of turning the light switch on and off manually, instead of using voice commands or an app.

[hansvm]

> It appears there's a misunderstanding here. When smart lightbulbs lose internet connection, they still function as regular lightbulbs.

That's a better way to install things, but it's not the only way that installation is done, and it's still excessive to need the internet for what could easily be a purely local operation. For device classes where "on/off" has historically been attached to the device itself (washer, stove, ...), it's more common than in the lightbulb case for the internet connection to be mandatory.

Pushing closer to tin-foil hat territory, when your lights can be controlled over the internet by servers you don't own, that operator can do nearly whatever they'd like. A common pattern in other industries (ocean navigation apps, OB2 monitors, ...) is to require a server for no good reason, sell a device at full market value, and then later extort the customer into some sort of subscription/data-leak to continue using their own devices. It's common for the remote server to just shut down (startup goes out of business, manufacturer decides it'd be nice if you bought a new device, ...). Less likely but worse if it happened, the server operator _could_ trigger epilectic seizures or whatever.

[BoxFour]

> That's a better way to install things, but it's not the only way that installation is done

It’s by far the most common. Most people just buy Hue lightbulbs and an hub to use with their existing home setup, which involves physical light switches still being present.

> Pushing closer to tin-foil hat territory…

For lights specifically: If this is a threat model that worries you, it’s quite easy to setup something like home assistant.io with zigbee lights.

Your repudiation appears to be about centralized services, not about “smart” devices as a general concept.

[hansvm]

I think we're talking past each other a bit. The person you first responded to said that some problems exist, you said they don't, and I expanded a bit, agreeing with them.

Your position seems actually to be though that you can make smart devices acceptable if you're careful enough. That's probably true. However, I've seen the code people write, and I still personally wouldn't want a smart oven or any number of other such devices.

> Your repudiation appears to be about centralized services, not about “smart” devices as a general concept.

There's some truth to that. I focused on that point just because smart devices usually have a markup for their extra features, even though you don't really own the product, manufacturers have a habit of shutting them down, and most of the "smarts" are just figuring out how to generate extra ad revenue. That's not _all_ smart devices, but it's a lot of them. Contrast that with a centralized service like Overleaf, where I know I'm renting server time and disk space, and it's obvious what the threat models are (they can read anything sensitive, and I should probably make sure everything is backed up). Selling living room audio recordings on a device you paid more than fair market value for just feels a bit slimy, and I thought that might be a good touchpoint to illustrate what sorts of things can go wrong.

[inferiorhuman]

I'm not offended by you enjoying a feature, but I see that feature as a net negative. An R&D budget is more or less a zero sum situation. Money spent developing temperature sensors is money not spent on something else.

Fridge quality today is lousy enough (and warranties + return policies have all but evaporated) that spending money elsewhere would've been a big win. Even if we're not talking reliability, I had to replace a 20 year old fridge with split shelves only to find that this feature is now only available on more expensive, less efficient, less reliable bottom freezer units. Another modern fridge I've dealt with recently (Whirlpool) was just breathtakingly poorly built. It worked, but imagine if Whirlopool had sunk money into QA instead of gizmos.

Presumably your built-in temp sensors are WiFi or bluetooth. What are the chances that's gonna get updated when the next WiFi standard starts to become more common? Will it support the next version of WPA? What happens when the power goes out?

When I had to diagnose a dying LG fridge, I got some cheapie ($20ish) RF temp probes. They're battery powered, the included receiver can set an alarm, and a cheapie software defined radio kit can decode the signals on Linux. I run rechargeable NiCad batteries in them and have moved them from the LG to a couple fridges since. Hands down a way better (and more robust) experience than anything a fridge manufacturer could dream up.

Door open alarms are comparatively much simpler and a good example of a feature with a much smaller inherent downside.

BTW, I'd also add that folks (not necessarily you) championing flashy features are forgetting just how dead nuts simple fridges used be. To the best of my knowledge my current fridge (Frigidaire / Electrolux) has a mechanical timer for the defrost cycle. The LG uses electronic controls to eek out a bit more efficiency. I know which one I'd rather replace.

[kevin_thibedeau]

It isn't zero sum. They could simply add the gizmos and keep the important bits the same as their refined, reliable 20 year old designs. What they're actually doing is actively spending engineering resources to enshittify what was already perfected to boost profit.

[inferiorhuman]

For better or for worse, no, they could not have kept the working bits the same. The US EPA has progressively tightened restrictions both on what refrigerants can be used and how much electricity can be used. Instead of focusing on making that reliable, they're focusing on making it cheap so they can add gizmos.

Besides, if you've got to replace a fridge every five years now that's a win for the appliance manufacturers. Although, really, who offers anything more than a 1 year warranty anymore? That should tell you how long these things are expected to last.

[darreninthenet]

Depends where you are... EU requires a minimum two year warranty for example

[inferiorhuman]

Yeah there are no such protections stateside. Consumer companies took one look at the pandemic and saw dollar signs. With big appliances you'll get a flat one year warranty pretty much across the board even with European brands like Bosch. Big box stores tightened up their return policies as well. You've now got 48 hours (down from a month or two prior) to return big appliances. Extended warranties with any purchase used to be a mainstay of American credit cards, and that's largely gone.

[bell-cot]

True, and if you're stuck with a fridge that's hooked to the internet regardless...

OTOH, this item's context is "KISS, because the complex products are hot & steamy piles of failure modes, all eager to show themselves off". From that PoV, the alarm features kinda sound like yet more things to break down.