Hacker News Reader

Comment by smt88 | Hacker News Reader

> Technology has advanced to the point where they can get some components to fail soon after warranty period expired with decent accuracy. Power unit failure tends to be a common one.

Unless the hardware is designed to be damaged (or restricted) by software, there's no way this is true. Technology can be great, but you can't "program" hardware to die on a certain date without knowing how it will be used, how long it will be used, what temperatures it will operate in, etc.

And you're not even considering that the manufacturer has no idea when the hardware will be sold and therefore doesn't even know the date that the hardware "should" fail.

Unless you have some evidence that this is both possible and happening, it's just a conspiracy theory.

View on HN

It could be from the standpoint of thermal design and having a reasonable idea of how long components will last at the temperatures they will typically run at. Realistically, all you need is a capacitor or two to fail in the ballpark to achieve that end with most customers. Granted, one wouldn't be able to to say 'after exactly X years of expected usage it will fail' but more likely be able to say 'after no more than X years of expected usage most units should fail'. Some heavy users would have theirs fail early and the company might have to correct via a warranty claim but that could easily be more than made up for by the masses of other customers that arrive for replacements roughly on schedule.

One doesn't even need to be that devious: just make 'consumables' like batteries and flash drives not easily replaceable nor easily sourced. Then just sit back and wait for them to start failing. As long as most manufacturers did this, you wouldn't have to worry about losing business to your competitors due to the practice. Pretty sure you can put a check mark next to most phone and laptop manufacturers on that front these days.

This isn't unique to tech. A couple of real world examples of planned obsolescence via designed-in failure modes: incandescent lightbulb filaments were designed to fail ~1000 hours of use (even though they could easily have lasted far longer) and most clothes washers are designed to fail after ~10 years (IIRC, manufacturers design a seal in the tub to slowly disintegrate via your laundry detergent.)

> One doesn't even need to be that devious: just make 'consumables' like batteries and flash drives not easily replaceable nor easily sourced. Then just sit back and wait for them to start failing. As long as most manufacturers did this, you wouldn't have to worry about losing business to your competitors due to the practice.

Based on the lack of profits from selling hardware in the first place, it makes more sense to me that these lower quality products are made with tradeoffs that result in a lower price for their target audience. The grand conspiracy idea, even if it were true, clearly did not pan out considering the loss in market value for basically every laptop brand other than Apple.

> This isn't unique to tech. A couple of real world examples of planned obsolescence via designed-in failure modes: incandescent lightbulb filaments were designed to fail ~1000 hours of use (even though they could easily have lasted far longer) and most clothes washers are designed to fail after ~10 years (IIRC, manufacturers design a seal in the tub to slowly disintegrate via your laundry detergent.)

Source? Unless there was only 1 or 2 washing machine manufacturers, it seems kind of trivial for a competing company to sell the tub with a better seal with a longer warranty. Once again, it is probably lower quality products aimed for people who want to pay lower prices. The speed queen or Miele washer with 5 year warranty is available for those who want to spend for it.

Personally, I would rather buy a washing machine that costs half as much from Costco which comes with a 4 year warranty, and trash it when it breaks. If another manufacturer wanted to sell a washer with a 5 year warranty that cost the same, or a 10 year warranty that cost less than double, I would be game for that.

The same situation for the light bulb. If there are competing manufacturers, people would rather pay more to not have to change light bulbs. Although, this depends on how easy it is to change the bulb. For a parking lot light needing a truck with a boom, I would spend double to get a quality bulb with a better warranty. For a house bulb I can change with a step ladder, I would rather pay $2 and gamble on it lasting longer than a $4 bulb, because even if it does fail after 1 year instead of 2 years, I am not going to spend my time warrantying a $4 item.

Light bulbs lifetime was open collusion because longer-lifetime lightbulbs are less efficient electrically. Lifetime electricity cost for incandescents is much more than purchase cost, so buying longer-lived bulbs costs you money.

I do not see how that is collusion in the usual sense of the world?

That is simply consumers opting for the better value, making the sale of longer lifetime bulbs an uncompetitive option.

It was the manufacturers that agreed not to sell non-standard 1000-hr bulbs, not consumers who stopped buying.

> clothes washers are designed to fail after ~10 years (IIRC, manufacturers design a seal in the tub to slowly disintegrate via your laundry detergent.)

Worse :( Seal can be easily replaced. They cast Drum Support/Spider part from specially selected alloy with curious property of easily Dissolving in washing powder :)

https://www.google.com/search?tbm=isch&q=corroded+spider+was... https://www.automaticwasher.org/cgi-bin/TD/TD-VIEWTHREAD.cgi...

Drum spider corrodes and falls apart, incidentally it is often the Only internal part not covered with anti corrosion paint :), and drum itself is obviously stainless. For comparison something like Miele is all stainless.

Audi (VW group) 2.0 TDI BLB Engine is made to fail at ~200K no matter what you do, from oil pump to head micro fractures. Crankshaft sprocket at 200K http://lh6.ggpht.com/_m_33vQTtsxM/SyuSlFIeSXI/AAAAAAAAG9o/rt... http://lh3.ggpht.com/_m_33vQTtsxM/SyuSn5YALLI/AAAAAAAAG9s/ul... http://lh3.ggpht.com/_m_33vQTtsxM/SyuSp1yUs4I/AAAAAAAAG9w/mO... VW answer: this part is non serviceable, whole crankshaft needs to be replaced :D Why is it made out of putty while smaller sprocket (the one taking more force) is still 'working'? Why is it designed not to be replaceable? Good 'old timer/not by the book' mechanic is still able to replace it (heat whole crankshaft, use big hammer). 200K is usually first crankshaft turn/grind, engine block itself is good for >500K easily.

> incandescent lightbulb filaments were designed to fail ~1000 hours of use (even though they could easily have lasted far longer)

This is repeated a lot but not correct. All the "forever incandescent" bulbs have one thing in common: they're dim, and they have a very red color temperature. They live long, because their filaments are much colder. If you want whiter light, and more of it (because lower filament temperatures are less efficient, since they emit even more infrared, as the entire spectrum is more red-shifted and flatter), the only way with incandescent is a higher filament temperature, which inherently lowers lamp lifetime.

What changed, besides lifetime, when lamps went from around 2000-3000 hours to 1000 hours is that they got way brighter and more white. Which is what customers wanted.

although it certainly is true these days, that the average LED based bulb is pushing those LEDs hard, which incidentally also makes them less efficient, rather than just using more of them at the same wattage.

Check out the dubai lamp to see a lamp designed to be efficient and last:

https://www.youtube.com/watch?v=klaJqofCsu4

The Dubai Lamp is not that special. There are similar lamps with only slightly more wattage, but also slightly more lumen. Osram Professional LED Retrofit Classic, or somesuch for instance. In Europe. Don't know about the Americas, Asia...

I googled that model, it doesn't really seem to have all that many filaments for its wattage rating, so I doubt it's underdriving those LEDs like the dubai lamp. The special thing about it isn't the lumen/wattage, but the longevity by reducing the amount of stress on the LEDs

When this made the rounds first a while ago, I wondered what the fuss was about and compared. At that time I had only the 'candles' with E14 socket, non-dimmable, clear, at 2.5W, 230V/50Hz:

[1] https://www.osram.com/ecat/PARATHOM%20Retrofit%20CLASSIC%20B...

which seemed comparable, though running at about double wattage.

Anyway, they last and last, and I'm perfectly happy with their light.

Meanwhile I upgraded everything else also, and have a few of these

[2] https://www.osram.com/ecat/PARATHOM%20Retrofit%20CLASSIC%20A...

also non-dimmable and clear, for lamps with a single E24 socket, which should be bright, at 7.5W & 10.0W, 230V/50Hz.

Still very happy with the light, but can't say much about lasting, because only about 2 to 3 years in use, while the candles are now 5 to 6 years old.

All 2700k btw. I like it that way.

edit: I should add, before that I followed the fad with the CFL stuff, none of which lasted longer than 6 months,

made ugly light, and that not even instantly(warm up), sometimes with noisy chirping.

So for me it looks like I should have skipped the CFL phase until these were available.

To be honest I wouldn't want to light living spaces with CRI 80 LEDs, no matter how long they last or how efficient they are.

Yah, Yah. I hear that so often. Maybe I'm colorblind, and you are all tetrachromates, but I don't see any color-misrepresentations here. And I've tested that.

But it has been proven formerly that the https://en.wikipedia.org/wiki/Phoebus_cartel existed.

OTOH, given how all consumer electronics contain complex microcontrollers running some RTOS and usually a buffered RTC, would it be inconceivable to have some periodic check running there, and after 2 years of time, changing some voltage on some path to ever so slightly higher, spinning some fan just a little bit slower, doing other weird pwm stuff, and have it cooked? Who would ever know with obfuscated firmware, deeply embedded and encrypted?

I'd say it would be less complicated to implement than the stuff which caused Diesel-Gate.

>Unless you have some evidence that this is both possible and happening, it's just a conspiracy theory.

It's quite easy and I think most manufacturers of electronic devices, electric appliances and even cars are doing it.

You just have to calculate what the average working time during the warranty period and adjust the quality of materials and the build quality so the MTBF is just a bit higher than that time.

> even cars are doing it

Cars are not doing it. Like I said: provide a source that isn't your own suspicions and anecdata.

I am a shareholder of an auto insurance company, and our actuaries would know if cars were doing it. There is a lot of very high-quality data about vehicle reliability, and major trends are very visible.

Car companies are often stupid and/or evil, but they can't (and don't) precisely plan obsolescence. Car usage is a huge bell curve, with the average American driving ~15k miles/year, but large groups of people driving far less or far more. You couldn't possibly plan a car to fail at X years.

Car manufacturers usually set the warranty as 36 months/36,000 miles, for example. I'm not claiming that this is the case, but it'd be far easier to plan for X miles than years.

>Unless you have some evidence that this is both possible and happening, it's just a conspiracy theory.

we are close to 2022, this shouldn't really be a conspiracy theory anymore

As some have mentioned Lightbulb is one of those prime examples. Where vast majority of LED lightbulb offer say 10 years of warranty, and yet because manufactures know most people dont bother sending them in, they cheap out on quality. Most LED ( if not all LED ) dont fail at all, it is the controller that fails, it is expensive and makes the difference. As long as it last 3-4 years, no one is going to complain. People will attribute this failure to bad luck / QA and just buy another one.

As well as other mention of capacitor. One easiest way is to get have Power Unit fail after 3-5 years, which seems long enough for Consumer Electronics. Most dont go and figure out it was X failure and just buy a new one. Synology J series NAS power unit is an example of this.

Consumer Electronics has been doing this in China for over a decade. There are reason why things are cheaper. And most internet user are still stuck with "Spec" comparison as if spec means everything.

>And you're not even considering that the manufacturer has no idea when the hardware will be sold and therefore doesn't even know the date that the hardware "should" fail.

Electronics get used when they are sold. MTBF since usage in many cases can be extremely predictable.

There is another point to look at it, a lot of consumer actually prefer to have CE fail within reasonable time so they get an excuse to buy something new. The market for some people like me who thinks something should last at least 10 - 20 years is ridiculously tiny. Although the buy for life internet following is certainly helping a bit.

> As well as other mention of capacitor. One easiest way is to get have Power Unit fail after 3-5 years, which seems long enough for Consumer Electronics.

You're moving the goalposts here. You said that:

> they can get some components to fail soon after warranty period expired with decent accuracy

But then you specifically mention a window of 2 years. From a business perspective, 2 years is not "decent accuracy".

I am pointing at normal circumstance with example for Synology, which in itself is a decent brand. Power unit is also the easiest to game when the devices is constantly on. There are cases where the accuracy is in months. Digital TV STB sold that last only 9-12 months with a 6 months warranty. And all it takes is using low quality components and capacitor in a special batch.

These dont ever reach the shore of US or EU unless you order them via Aliexpress. Consumer protection and expectation sets things apart. But doesn't mean vendors dont do it.

I also gave the LED light bulb example where evidence are given with respect to warranty and lifetime usage.

And no one implied that all of these product, every single batch of it will specifically failed at the same time. Which I think is what you are getting at. My original post only mention it is technically possible, and in many cases actively done so.

It's not really a conspiracy, just the natural outcome of design optimization - any part that still works reliably at the end of the design lifespan is a candidate for cost cutting in the next iteration.

> any part that still works reliably at the end of the design lifespan is a candidate for cost cutting in the next iteration

This assumes a positive correlation between quality and price for parts. That isn't always the case. There's no reason a better part will be more expensive.

And again, you're ignoring my primary argument: "lifespan" is not measurable in calendar years for electronics. Some people I know use their laptops 360 hrs/mo, and some people (like me) only use them when traveling, ~20 hrs/mo.

After a calendar year of that, my laptop has only 240 hrs of life lost, while another laptop might have 4,000+ hrs of life lost.

Please provide a source about how a hardware manufacturer (without the aid of software) could plan for a component to fail after either 240 hrs of usage or 4,000 hrs of usage.

It’s fairly easy you calculate the MTTF when you design anything. That said it’s probably not as malicious as people might think their goal isn’t to artificially lower the MTTF to just beyond the warranty period but rather that they don’t put any effort in extending the MTTF beyond that.

And really it’s not that companies in the past did that much better, it’s just that in the past getting replacement parts was easier and you had fewer integrated components. Today everything is pretty much in a single package even basic electric circuits like a bridge rectifier. So things got smaller and are more likely to fail and they are harder to replace.

> It’s fairly easy you calculate the MTTF when you design anything.

Where did I disagree with this?

What I'm saying is that MTTF is measured in usage time, whereas warranties are measured in calendar time.

Cycles under normal usage conditions aren’t that hard to calculate.

The range of "normal" is massive for things like cars and laptops.

Again, provide a source please. Otherwise this is just unprovable speculation on your part for something that is theoretically difficult in the first place.

Reply on news.ycombinator.com