Hacker News Reader

Comment by nexuist | Hacker News Reader

Kind of an interesting dilemma. Charge full price, only rich people can afford it, poor people die. Charge a subscription, poor person affords it for a while, bank account gets overdrafted, poor person dies. On the other hand, charge a subscription, poor person affords a safety feature they previously couldn't use, saves their life. Isn't that the more ethical option? If the full price is $800 then they're actually selling the vest to you at a 50% discount and hoping to recoup the costs through the subscription. You're getting a safety product for 50% off and you don't even have to pay back the other 50% if you don't think you'll need it. Hm.

View on HN

>Isn't that the more ethical option?

No. As a civil engineer that routinely works with and designs safety-critcal systems, and understands the concept of "public trust" and personal liability as Engineer of Record, whoever designed this is an absolute psychopath.

Exactly. That's why "software engineering" will never have the equivalent of a PE. Terribly inconvenient to "move fast and break things" when you're obligated to give a shit about the collateral damage when something breaks.

This is not true at all. There are entire, large sections of the industry devoted to creating safety-critical software. These areas of the industry are well disciplined and highly regulated. This software is designed in such a way to make collateral damage impossible in all but the most exceptional scenarios. This popular silly web programming stuff is truly only the tip of the software engineering iceberg.

No, it's completely true. There are relatively tiny, quite niche sections of the industry devoted to creating safety-critical software in a systematic, engineered manner. I've been part of it. My company has more people writing Java/Cobol business apps than all of NASA has writing true high-reliability code. Even then, while folks writing mission critical code might be 'disciplined', there is no equivalent of the NSPE regulating what someone needs to know to do this, no universally agreed upon standards for what constitutes "safety-critical" or even consensus metrics for how to assess whether something is or not. There are literally hundreds of different guidelines for "how to engineer safe software", but no canonical guide. Even stuff like "can you use C to write software 'in such a way to make collateral damage impossible in all but the most exceptional scenarios'" is a matter of furious debate; some say yes, some say no.

Not just that, but in PE land, the Engineer of Record is PERSONALLY liable for what they stamp, and no corporation can really stand in the way and protect them from that liability.

The automotive industry has standards that govern the systematic development of safety-critical systems (hardware and software). The most significant is probably ISO 26262, which defines an “Automotive Safety Integrity Level” and methods for ensuring the safety and reliability of components at a given integrity level. There are also standards like MISRA C, which are basically defined code quality and software development guidelines to reduce bugs in critical software.

So...exactly my point? Each niche has it's own idea of what the standard is.

> whoever designed this is an absolute psychopath.

Yes, if the product were to refuse to protect you after you started riding, that would fit your description, and I would agree. I think it doesn't do that though.

From an engineering perspective, what are the safety risks here, exactly? Did you review the design spec or read the product page?

"3. If I suspend/pause my subscription and forget to reactivate it, will the vest still detect a crash and inflate?

"Answer: No, because in the first place you won’t be able to turn it on into ride-ready status before your ride. It will not “trick” you into thinking it’s active when it is not, because the LED indicators will warn you that it’s not active. If you ignore the LED indicators warning you that the airbag isn’t active, you can’t expect it to work when you’re not actively subscribed."

https://www.klim.com/Ai-1-Airbag-Vest-3046-000

Given that description, what are the failure modes you envision that are "psychopathic" and would endager the rider and/or erode public trust?

>If you ignore the LED indicators warning you that the airbag isn’t active, you can’t expect it to work when you’re not actively subscribed.

On the things I work on, where public trust is at stake, there are typically several layers of redundancy in not just the design, but also the regulation, all designed to prevent human error from occurring at a later date.

Here's a very simple example for a very common and low key activity: In my county, wastewater treatment regulation is based on the assumption that 100 feet of horizontal separation provides sufficient treatment and protection between a groundwater well supplying potable water, and a septic system. Now, both of those require permits to install which requires a site plan and coordinates, etc. (Check number one.) It is especially illegal for well drillers to drill a well without filing the permits and record with the county. (Check number two.) You cannot permit either a well or a septic system within 50 feet of a property line. (Check number three, which helps to ensure 100-feet between systems even if someone makes a mistake about what is occurring on an adjacent property.) When permitting a septic system, you must dig a deep hole to investigate and make sure that there is sufficient distance to groundwater through the soil to provide treatment before reaching groundwater. (Check 3, this provides redundancy in case there is some form of short circuiting, or a well get's drilled nearby anyway). The soil type is lab-verified to ensure it does not drain too quickly, compromising the treatment before effluent would hit groundwater. (Check 4: redundancy) In many cases, a monitoring well is required with a minimum of 6 inspections to further ensure that groundwater is sufficiently deep year round to avoid compromise. (check 5). Many of these steps require BOTH a licensed engineer and a county representative to be physically present. Etc. etc. etc. And this is an example with compartively VERY FEW regulations and checks to ensure that someone nearby you doesn't get sick.

>Did you review the design spec or read the product page?

Yes, I also ride motorcycles and have reviewed several motorcycle airbag options.

>... "If you ignore the LED indicators warning you that the airbag isn’t active, you can’t expect it to work when you’re not actively subscribed." ... Given that description, what are the failure modes you envision that are "psychopathic" and would endanger the rider and/or erode public trust?

1.) Human error to notice, identify, and correctly interpret an LED indicator light.

2.) Human error to properly correct an LED warning light indicator for a variety of reasons. ("I called them and gave them my new credit card so I should be ok, even though the light is on.")

3.) Inability to correct the situation without first continuing travel to an area with cell or internet service. (Date changed, subscription expired, have been traveling and haven't had reliable cell, internet, or mail for 2 months.)

4.) Hardware failures of a huge variety, which can exist in a system which has an inbuilt mechanism to DISABLE airbag.

5.) Programming errors and/or network errors.

6.) Hacking: The device is "connected and upgradeable". Enough said.

This whole thing is a terrible idea, to the point, where, I WOULD NOT TRUST the device even if I purchased the non-subscription version, lest it somehow fail after an unauthorized update, or revert to thinking I'm on a subscription service.

...The thing is, in engineering where public trust is involved, you simply DON'T DESIGN SAFETY SYSTEMS THIS WAY. You design them to be as hard to screw up as possible even in the face of substantial human error, or negligence, or malicious intent.

Would you feel ok entering a building where the fire exit required the building owner pay a subscription service to unlock?

Thanks for the detailed reply. So it seems like your objection has nothing to do with the payment model, right? All of these are just generic engineering concerns with any product at all that has the ability to electronically enable or disable a safety feature.

Why are you concerned about and this particular product, as opposed to say cars, which affect orders of magnitude more people? Modern cars have many safety systems that can be digitally activated and deactivated, including the steering, brakes, throttle, and airbags, by design. The ECU alone in any vehicle, car or motorcycle, is hackable and could kill you (did in Toyota's case). All vehicles have LED lights to indicate safety critical problems. Nowadays, it's more commonly Android-driven LCD panels with many, many more failure points than an LED.

I mean, I hear you that electronic systems have many failure modes. But electronic control is more than routine today, it's utterly ubiquitous, so what's the cause for such extreme concern here that words like "psychopath" are justified?

Tesla alone is putting more electronically controllable cars on the road than all highway motorcycles combined in the US per year. Should we even talk about the failure modes possible with a Tesla or other self-driving car?

> Would you feel ok entering a building where the fire exit required the building owner pay a subscription service to unlock?

If I had to manually power up the entry way, and the building warned me before I entered about the subscription status, it wouldn't bother me that much. Please do at least try to be fair to the perspective of the business Vice has misrepresented here.

Before I reply, I want to recognize that as the actual licensed engineer, you have more of a say in this than I do, a non licensed software developer, regardless I would like to continue the conversation to better understand why I am wrong.

Your core argument seems to ride on the idea that public trust is paramount. I can definitely see this on projects such as bridges, buildings, wells etc. I struggle to see it on a private consumer good such as an airbag vest. If I want to cross a river I'm forced to drive across a bridge, and the bridge doesn't come with an instruction manual. It makes sense to engineer for the lowest common denominator here so that the functionality is predictable.

But I'm not forced to use the airbag vest. Every issue you listed is a problem with almost all motor vehicles. On my car dashboard the LED that pops up to indicate low tire pressure is tiny and the label so abstract I had to look in up in the drivers manual to figure out what it meant. In many situations low tire pressure means you could be seconds away from losing control of the vehicle, yet the engineers trusted that dashboard LED would be safe enough for consumers to understand. This is also true for ABS, traction control, stability assist, brake failure, and whatever else gets its own LED.

Even disregarding the subscription service, it still makes sense to have a device that shows a status LED to confirm the system works. What if the battery died? What if some wire internally disconnected? What if some hardware component failed a startup check? An electronic system can alert you to failure conditions far better than a mechanical one.

> 4.) Hardware failures of a huge variety, which can exist in a system which has an inbuilt mechanism to DISABLE airbag.

On any modern passenger car the passenger airbag is disabled if a weight sensor in the seat isn't tripped. These systems already exist and are deployed en masse.

> 1.) Human error to notice, identify, and correctly interpret an LED indicator light.

> 2.) Human error to properly correct an LED warning light indicator for a variety of reasons. ("I called them and gave them my new credit card so I should be ok, even though the light is on.")

> 3.) Inability to correct the situation without first continuing travel to an area with cell or internet service. (Date changed, subscription expired, have been traveling and haven't had reliable cell, internet, or mail for 2 months.)

> 5.) Programming errors and/or network errors.

From the site: https://www.klim.com/Ai-1-Airbag-Vest-3046-000

4. If I’m subscribed monthly and miss a payment or my credit card expires, will the airbag suddenly stop working?

Answer: No. In&motion will reach out with a 30-day warning prior to your payment method expiration. After expiration or missed/forgotten payment, In&motion gives you a 30-day grace period during which the airbag functions completely the same. You will receive notifications about the missed payment. After the 30-day grace period, the airbag will stop detecting crashes until payment is resumed. You will not be able to turn on your airbag vest into ride-ready status after the 30-day grace period, so you won’t unknowingly ride with a non-functioning vest, unless you choose to ignore the LED indicators warning that it isn’t active.

5. Would In&motion turn off my vest mid-ride if I pass the 30-day grace period?

Answer: No. In&motion will not turn off your vest mid-ride for any reason, even if you’re at the end of the 30-day grace period and it elapses during your ride. The only way for it to turn off during a ride is if you didn’t charge the In&box and the battery dies (LEDs indicate battery level, which you see when you turn it on) or if you manually turn it off during your ride.

> Would you feel ok entering a building where the fire exit required the building owner pay a subscription service to unlock?

This is a false premise because fire exists are required by law and factored into the cost of the building. There is no legal situation where you can skimp out on fire exit costs. Also, a motorcycle isn't a shared space. Whether another rider chooses to wear protection or not has no impact on me as a car driver.

Thank you.

The ability for the vest to check if your still paying the subscription costs more money to build than an always working vest.

If it actually costs 800$ then you can give people the option to make 40 payments of 20$ a month easily like how many people buy iPhones and sofas.

This is a great point. The IoT subscription model can be avoided by using a more traditional rental model. I don't think anyone would have qualms over financing plans about this, but the IoT angle makes it more controversial.

Not a dilemma, it's just shitty.

Where does the $800 figure come from? As far as we know, it's just a number they made up. I have strong doubts they would be able to sell the vest at that price if only the "full price" option were available.

For context, the price of a non-airbag motorcycle jacket is typically $150 for a basic one, $250-400 for a good value, and up to $1000 or more for a high end jacket.

Also, the cost of a major motorcycle injury is easily six figures before insurance.

Reply on news.ycombinator.com