This is no surprise. We are all learning together here.
There are any number of ways to foot gun yourself with programming languages. SQL injection attacks used to be a common gotcha, for example. But nowadays, you see it way less.
It’s similar here: there are ways to mitigate this and as we learn about other vectors we will learn how to patch them better as well. Before you know it, it will just become built into the models and libraries we use.
People seem to ignore the cost and accuracy aspects of a phone listening to you 24/7. At least with today’s constraints, it is highly unlikely to be happening.
First, the cost to transcribe audio is not free. It is computationally expensive. Any ad network or at scale service would not be able to afford it, especially in orgs where they are concerned about unit economics.
Secondly, the accuracy would be horrible. Most of the time, your phone is in your pocket and would pick up almost nothing. More over, it’s not like you are talking about anything of value to advertisers in most cases. Google is a money printing machine because people search with an intent to buy. The SNR of normal conversation is much much much lower. That makes the unit economics of doing this gets much worse.
Third, it would be pretty hard to not notice this was happening. Your phone would get hot, your battery would deplete very quickly, and you’d be using a lot of data. Moreover on iOS you could see the mic is being used and the OS would likely kill the app if it was using too many resources in the background.
So until we find an example of this actually happening, it’s not worth worrying about.
For all of these reasons, audio snooping is much more likely to be something done by wired, stationary devices that maybe have a decent amount of RAM + a fair bit of usually-idle processing capacity (to run the transcription model locally and just push the resulting text), and which are expected to draw a decent amount of power and use the Internet at vaguely-arbitrary times.
These are all points that were brought up in the article as to why voice recording is less useful than all of the other tracking mechanisms advertisers have available
While I think that audio recording is not a thing, your economic argument is not complete.
What if only the audio of "high value" targets is recorded. Meaning people who buy a lot of stuff. So it might be worthwhile to only record their sounds. Which will explain why random testing (usually with new/clean phones) is never successful in detecting a recording event.
I think this is a genuine concern for prominent people. Like if you are Mark Zuckerberg, there is material interest in a bad actor installing malware on his laptop. But for a random person where you get low value data that may or may not let you better target some low value ads? That is much harder to justify. Would have to reevaluate as things change and the cost of compute goes down.
I think there are plenty of movies that aren't remake-of-a-copy-of-a-sequel out there. I recall I liked Didi.
I also think a lot of people have a better experience at the movies. The movie gets watched straight through. It isn't abandoned, stopped to get snacks, or text, or take a phone call. If someone keeps falling asleep, you don't rewind 10 times. If the movie starts out too slow or subtle, you keep going instead of scrolling to find another movie. etc.
San Francisco uses RCV, and it’s not much better, maybe worse. Yes you get run off elections and more candidates. But now voters have to use strategy in how they vote and it’s complex to understand the implications. There’s a higher chance of winding up with unpopular candidates simply because nobody actually wanted their second or third choice candidates.
This reminds me of when I went in to a Patagonia store to repair a jacket with a “lifetime” warranty. Turns out they define lifetime as the “useful” lifetime of the product, which is a couple years. They refused to help and instead tried to sell me a new jacket.
I forget who owns Black Diamond, but they're kind of similar.
They haven't fully replaced the product, but what is cool is that they have a repair shop that has been doing free repairs for me. I've sent a very lightweight, very heavily used puffy jacket in twice for repairs at no charge.
Realistically I know that jacket isn't going to last forever, but I respect they are at least trying to help me extract as much life out of it as I can from a sustainability perspective.
It's using less material and less landfill, but I wonder if it really is more sustainable in the grand scheme of things, at the scale of clothing and similarly sized items. The additional round trip shipping and workshop operations (HVAC, lighting, commuting, etc.) could potentially exceed the footprint of just sending you a replacement right off the production line. Obviously there's a crossover point above which this couldn't possibly be (cars, etc.) but it's probably a very blurry line, and I wouldn't be surprised if some companies knowingly take the worse but ostensibly sustainable option, i.e. greenwashing, for the resulting brand loyalty and word of mouth advertising.
It absolutely is, but [using it for production of new items and using it for a repair shop] might take more resources than [just having the former and supplying some replacements]. What I'm saying is that we can't just compare consumption/waste of materials (which is obviously worse when doing replacement instead of repair) because there are also "overhead" resources required in order to offer repairs. Theoretically, in cases where replacements are better for the bottom line than repairs, it's due to using fewer resources, and the open question is how "green or dirty" those resources are.
If replacement is cheaper only because of geographic differences in wages, then we ought to repair. But if replacement is cheaper because of streamlining the use of nonrenewable electricity and so forth, then we ought to replace.
I sent back a down jacket which weighed practically nothing and packed down extremely small. However, harvesting the down is somewhat controversial and only recently has there been a movement to use ethically-sourced down feathers (I haven’t looked into the RDS standard. I’m sure it has problems, but hey, it’s a step in the right direction).
For normal fabric clothing, I think you are probably right. I do feel like the roundtrip in this case was worth it to get the most usage out of the feathers as possible (not to mention the 1000+ fill jackets like this are expensive).
Darn tough socks still honor their lifetime warranty no matter how long passes, though obviously no socks can last forever. Generally reading online you find people mentioning you should be reasonable about it.
I don't get it. Shouldn't it be the seller's obligation to give a reasonable lifetime estimate? Like, give me a five year warranty, if you want to advertise your socks last for five years of regular use. Don't pretend it's unlimited when it isn't.
>Shouldn't it be the seller's obligation to give a reasonable lifetime estimate?
Not sure how you define this or maintain it. These socks are guaranteed for 100 wears? Can't count wears. These socks last a year. Is that daily wear? One of 10 pairs? Only air dried? Was the user running daily marathons?
buncha people caught wind and purchase the product used/torn for pennies on the dollar, and send it in, in order to take advantage of the offer (and the retailer).
Statement still stands. The company can't afford lifetime because of this possibility. They should change the terms. They could say single owner lifetime or something like that.
Tilley hats as well. It was probably twenty years on, and both of ours fell apart enough to call about their lifetime (“put it in your will!”) warranty. Other than arguing that Tilley never made that model of hat, they sent us an equivalent without fuss.
By that standard we would have "lifetime" warranty on everything sold in The Nederlands, since by law we require warranty as long as you can reasonably expect a product to last.
My Tomtom GPS is like this. I have an older model. "Lifetime maps". For many years, plug it in, new map, download done.
Eventually I try to update it and it says "oh no, do you want to buy a map!?". I mean. What? Doesn't even cost anything to the company to keep on giving me free maps - well I guess it's lost revenue if that they could earn by dishonoring the agreement, which is what they did. Clearly meant to extract more money from me in map purchase or to buy another "lifetime" map.
I have another TomTom on my other vehicle (despite the shitty practices, their kit doesn't randomly crash like Garmin in my experience) which about every 2 days nags me about an update. So here I am, newer model is way too aggressive with updates all the time, old "lifetime map" model is a disaster.
What it is here, is there needs to be legislation that if a company uses "lifetime" or equivalent word in marketing, they are on the hook for life to honor that, with some prescribed action to make customers whole if they should want to drop it.
Now a good guy legend in this field, craftsman tools, for many decades in america people would buy craftsman from their sears knowing they could always go back easily and get a replacement. Sears in the day was like if Wal mart and amazon was the same company. An institution.
I've taken multiple 10 year old T-Shirts with holes through 10% of them in to the Patagonia store and they've let me walk out with new product off the rack.
I had a defective ATX psu cable and MSI support sent me a whole cables kit overnight. And recently a bought a Corsair case, the iCue controller had 2 defective ports and Corsair also sent me a replacement overnight.
My only "trick" with support is telling them upfront that I will leave a 5 stars review on amazon uppon successful resolution of the problem.
Not anymore. I had a pair of boots that one of the soles fell off of one day. They were about 20 years old but still in good shape except for the glue failure. I called up LL Bean and they said they had no record of the purchase (I didn’t have a receipt but I bought them directly from them). After I insisted I had bought from them they changed their tune to saying 20 years is long enough and I should know that glue on the soles of shoes fails after a while. I just wanted them to repair the boots but they refused, so I won’t be buying anything from them anymore.
Here was their old guarantee: "Our products are guaranteed to give 100% satisfaction in every way. Return anything purchased from us at any time if it proves otherwise. We do not want you to have anything from L.L. Bean that is not completely satisfactory."
If they had just worn out, that's fine, but they weren't at all. I only use them maybe 5-10 times per year, so the sole still had tons of life in it. The problem is that they didn't sew the sole onto the boot upper and the glue they used just lets go after a while.
I don't think that the soles should just fall off your boots one day while you're hiking, so no, I was not completely satisfied and I would like them to glue the soles back on for me.
Yes, it seems cruel to push young doctors to the limit like this. But I’ll offer a counterpoint:
In an emergency, you want doctors who are used to making decisions under stress and who are aware of their impaired decision making abilities when tired. This is a rite of passage that means in a true emergency where they have to be making good decisions without adequate resources they can do so. You see a similar tactic when training military recruits.
I don't know if you need the duration that residents undergo to get that stress training benefit. But I also don't think we have enough training throughput of doctors to prevent the situation where a tired doctor has to handle an emergency.
It's also quite frustrating the money is not there for the work put in and the personal sacrifice.
But doctors do not end up "aware of their impaired decision making abilities when tired" after that grueling period. They emerge convinced that they are used to being tired, able to work long hours and generally an exception to the "decision making abilities are impaired" thing.
The situation is normalized in their heads, they lived it so long that they see it as normal.
And with military recruits, they are made to sleep a little so that they are easier to coerce and mold. Then not being able to think is a feature.
For models of this size, the code used to train them is going to be very custom to the architecture/cluster they are built on. It would be almost useless to anybody outside of Meta. The dataset would be more a lot more interesting, as it would at the very least show everybody how they got it to behave in certain ways.
People take these kinds of things way too literally. There is no golden solution here. What gets repeated over and over continues to be true: teams should choose a system that works for them. And ideally that system is measurable, so the team can evaluate progress, improve its own performance, and align itself better with other teams and the business.
But in terms of scrum and points here's my take:
I've seen points work on some teams and not work so well on other teams. It's imperfect, but if you just accept that, you can make it work quite well.
The reason it's helpful to estimate complexity as opposed to time is that people with different experience levels would give different estimates based on their abilities. Complexity allows you to rally around a common understanding of a solution regardless of how fast one team member might be able to complete it versus another.
Does complexity have some relationship to time? Absolutely. Everybody knows this. That doesn't mean that we should be using time instead.
So how can a team estimate accurately? You will hear from some people that their estimates were wildly off or that it's impossible to estimate a project or they felt pressure to under-estimate. If your estimate is too broad, you need to do the mental work of breaking it down into smaller chunks that are easier to estimate. If you feel under pressure to ship on an unrealistic schedule, that's not a points/scrum problem. But the "it's done when it's done" is also not realistic either.
The idea that the estimate has to be 100% spot on is also not true. Again, it's imperfect and that is ok. But you'll find that the better a team knows their codebase and knows the product, the better they'll get over time at estimating. But if the work is too vague, the team should push back until they have enough information to more accurately break things down. This process makes for better software, especially when the team does it together.
Another missing aspect I see a lot is having a feedback mechanism. If you as a team are discussing why a task took longer than the estimate, or track metrics over time, you can all get together and figure out where problems on the team are. For example: maybe there are too many bugs that are hindering product work? Why? Maybe you're moving too fast vis-a-vis the expected quality bar. Some sort of feedback mechanism (e.g. retros) is crucial - the team as a whole should aim to deliver what it says it would and understand why it couldn't.
The whole point of these things is that as a team you can deliver consistently not more speedily. Consistency comes before speed. The other important thing is having a way to continually improve. You want to use each sprint as a way to measure the team so it can get better.
When I've seen teams that did this well, they were dramatically more productive than the teams that didn't do it well.
> IMO, the solution is apprenticeships/probationary offers
Most companies would love this, especially startups. The problem is that desirable candidates do not love this arrangement - there's an opportunity cost and risk for the candidate here and if you're in demand you can get a solid offer from a company who isn't trying to hedge.
I’m not saying I would trust getting into a Waymo now in those conditions, but I also wouldn’t assume the same things that are difficult for humans will be difficult for self driving. I’m optimistic these hurdles can be overcome.
Oh I'm also optimistic they can be overcome, I'm just less optimistic on the timeline. I'll be pleasantly surprised if self driving can handle winter a few decades from now.
There are any number of ways to foot gun yourself with programming languages. SQL injection attacks used to be a common gotcha, for example. But nowadays, you see it way less.
It’s similar here: there are ways to mitigate this and as we learn about other vectors we will learn how to patch them better as well. Before you know it, it will just become built into the models and libraries we use.
In the mean time, enjoy being the guinea pig.