It just goes to show data cleaning is hard. It’s true for statistics and it’s probably just as true for LLMs.
Techno fascism never looked so bland
I installed the Wendy's app a while back, and by optimizing my selections, all of my orders since then have cost less than half of menu prices on average. For example, get a $3 item for $1 with any other purchase (so you tack on a $1 frosty), or buy a $6 item get another free. Taco Bell's app deals aren't quite as deep but you get the point. When tons of people are placing orders this way, menu prices must creep up to compensate.
It's weird, but it's actually an incredible way for the more well-off to subsidize the less well-off (at the risk of enticing some folks, who have the means but love a deal, to be penny-wise pound-foolish at everyone's expense including their own). At the end of the day, more people get fed at the price point they can deal with.
If no one did this, wouldn't that just mean that the price discrepancy kept rising and rising?
Now introduce the middle. The portion of them paying with coupons is anywhere from 0% to 100%. The restaurant makes massive profits from them at 0%, and makes no profits from them (but covers all costs) at 100%. Let's say it starts near 0% because initially nobody knows about coupons and over time rises near 100% once the information has spread around. Profits decrease so something must be done to counter that. Shallower discounts (say, 25% off instead of 50% off) is a possibility, keeping the discrepancy tight, as you suggest. But increasing menu prices, while leaving the huge discrepancy, is also a possibility!
Either method will get the profits to the desired level. But only one method -- maintaining the large discrepancy -- sustains the goal of maximizing the number of poor people fed. Therefore, a restaurant with that goal will choose that method; a restaurant without that goal simply won't issue coupons. The discrepancy, if one exists, is therefore constant regardless of how many penny-wise pound-foolish people are in the middle.
As for app vs website, I agree, website would be similarly good. This could be said about tons of apps.
Of course all of that is just the icing on top of the data harvesting cake
https://www.youtube.com/watch?v=LLIj3pXOKjs
and came to the conclusion that order accuracy is the best if you use the app.
Saves having to wait on a line at the drive thru. Or on a line inside the store. Nope, can't do that: they don't take orders at the cashier, so you have to use the kiosks which take even longer.
So I use the app to order while at home, drive to the restaurant and grab my food and leave.
Apps are used rather than PWA websites because most users find it difficult to save a mobile website to their homepage and mobile web push notifications etc add extra friction compared to native apps.
In contrast I struggle to find apps installed on my iPad because the icons look all the same. Apple has a leg up on Android but for me Apple's icons are mainly forgettable or meaningless and most icons from third parties are a forgettable stylized letter, forgettable anime character, or abstract icon. The colors on the default background often obscure the edges of some icons so I find it hard to spot even icons I use a lot.
As a result I hide as many Apple icons as I can (What's the difference between the App Store, Apple Store, and iTunes store?) and avoid installing apps because each app I install makes it harder to find the ones I really use. The Taco Bell app would be brand destroying for me because I'd keep seeing it get in the way of finding the app I really need and would be popping up irrelevant and annoying notifications at all the wrong times -- you just don't want people associating your brand with petty annoyances.
There is no reason it needs to be a PWA. People had plain ordinary web sites to order food online a decade using cgi-bin and the equivalent before there were things like Angular and React.
> People had plain ordinary web sites to order food online a decade using cgi-bin and the equivalent before there were things like Angular and React.
This is true. I wonder if there is any difference in the $ amount of food ordered using iOS/Android apps now vs food ordered using cgi-bin then.
Back in the day people were much less into the e-commerce habit and not using mobile technology so they weren't ordering food on the go.
Another question is "What relationship to people have to fast food?"
I worked at a company that did geospatial analysis such as retail location selection. We had a theory that people chose fast food because they were on the way from point A to point B so the right way to think about it was not about the density of commercial or residential development in an area but rather about the density of trips that pass by a point. I was involved in a pilot project to use touchscreens to collect data at the POS just before the mobile age made it possible to collect trip data directly.
Thus my consumption of fast food is opportunistic: I eat at Taco Bell sometimes because it is in a neighborhood with a Wal-Mart, Gamestop, Petsmart, Staples, an illegal cannabis dispensary (not like I can't get better weed elsewhere), award-winning wine store, etc. I get hungry On most days I would go to the street taco stand on the other side of the parking lot. which has the best tacos I've seen outside Los Angeles but if it is Sunday maybe I go to the Bell. It's not like the scene in Demolition Man where Sylvester Stalone goes to a fancy dinner at Taco Bell because "all restaurants are Taco Bell" in the future.
If I am traveling maybe I am driving down the freeway and see a sign for a Burger King and stop. Maybe I walk out of the Oculus at the WTC site and see both a Chopt and a BK and, even though I have a BK gift card in my pocket, the line looks really long at the BK and I go to the Chopt because it is really fast food.
I can see that Taco Bell wants to develop a special relationship with me but I don't want to develop one with Taco Bell. Really I don't find it easy to install an app (until I broke my old iPad, my old iPad insisted that I log in with my Apple account password whenever I wanted to install an app despite the fingerprint scanner working just fine for everything else. I don't know my Apple account password because I keep it in a password manager and the app store was the one thing that would make me require to use it when I am on the go. When I bought a new iPad this cleared up. I go to the Bell maybe 4 times a year, it is just not worth having another app cluttering up my device making it harder to find the apps that really matter to me.
Speaking of which, we stopped going to fast food burger places because 75% of them can’t make things like “a hamburger with no cheese or any other toppings except ketchup”, or “a cheeseburger with no other toppings”.
I suspect an LLM that supports audio input would outperform most drive through window attendants.
Oddly, I’ve noticed there is no correlation between speaking english as a first language and being able to understand those orders.
(This is why a non-native speaker can do this job well, you don't need to know a lot of the language to parse orders like 芥兰牛肉)
I think this would lead to a lot of really frustrating errors in a food ordering context. In my experience LLM speech recognition is really prone to inserting or removing negation words, which can really shift the meaning. This is a problem in any speech recog use case but there's often sufficient context to understand when the error has occurred. In a list of what a customer does / doesn't want, there is no additional context to fill the gap.
"Cheeseburger with ketchup mustard onion no tomato" -> "Cheeseburger with ketchup mustard onion tomato" or the inversion
You can correct this to some degree with an order confirmation step but the error rate would mean a lot of customers need to make a correction, which adds time to the process and increases customer frustration.
At that point I'd rather the drive thru have a touchscreen and just let me input the order directly, versus having to argue with an LLM about which ingredients i want included or excluded.
Try:
“A hamburger with only ketchup.”
When you take hundreds of orders a day, circular communications won’t stick.
Once it led to a hamburger bun with just ketchup.
Well, they might be now since California passed a higher minimum wage for fast food workers.
> afford not to replace human jobs
There's practically no such thing as replacing human jobs; having a machine replace some of your /tasks/ (taking orders) lets you spend more time on the other tasks (preparing the orders) and makes you more productive.
PS: $18 an hour in the bay in 2024 might as well be under minimum wage. It’s unconscionably low if one wants to maintain any standard of living.
Depends on demand.
https://en.wikipedia.org/wiki/Jevons_paradox
> PS: $18 an hour in the bay in 2024 might as well be under minimum wage. It’s unconscionably low if one wants to maintain any standard of living.
That wasn't in 2024. It's $20 now.
100 miles from my home.
The other day I was at McD's and an elderly lady was stressin' it because she had no idea what to do. There were two cash registers but nobody staffing them and a disorganized group of people waiting for their food and not even a clear queue for the people behind the counter to see that somebody was waiting. She asked me what to do and I told her she could order at the register but a minute later I realized I was standing in front of the register waiting for my food and that probably the workers wouldn't see her. I stepped out of the way and made sure she was highly visible at the register and when I got my food I told the worker that this woman was waiting for her food and she seemed quite annoyed but I didn't feel I could take it for granted they were paying attention.
For our rural area with relatively low population, it makes sense. But even in the suburbs nearby, Chipotle, etc., are going to online-ordering only.
The elderly lady was probably doing what mom does when presented with a kiosk, pretending to be too dumb to click on a picture of the food they want.
Everything has an app. And pretty soon everything will have an “AI”.
So my conclusion it is just a data harvesting scheme.
FWIW, as one data point, this is what I've used it for a few times, and it worked without issue. (A co-working space has it next door and that made it a lot more time-efficient, avoiding the line and having to wait in store.)
Though I have run into the problem you've described with other services, where it rejects your request much later than it should. Like with AWS letting you configure a server for launch before informing you that you don't have permission to launch servers:
http://blog.tyrannyofthemouse.com/2016/02/some-of-my-geeky-t...
(2) The app gathers statistics. Marketing and sales are so addicted to massive amounts of all kinds of information they have trouble finding veins that are not collapsed and their coworkers all carry naloxone. In most organizations, the marketing and sales people make most product decisions and a mobile app is a (free) product distributed by the chain.
(3) Data collection. Not only can anonymized data be monetized, but selling data directly attributable to individuals is a major source of revenue for many organizations, and I doubt Taco Bell (or Yum! Brands in general) is not among them.
Price discrimination is about raising prices.
You think coupons and deals (e.g. the constant stream of BOGOs, late-night specials, et cetera) are new?
Any idea what the going rate is for the kind of data that could feasibly be collected from restaurant apps?
I don't have taco bell app, experience is from international McDonald's app
In my experience, they will send you discounts (BOGO) or free stuff when you're most likely to buy. One of my teammates and I both installed their mobile app. I would get free coupons for like nachos or a taco on the days that I regularly went, usually in the mornings and the coupons were only good for about 24 hours. Conversely, my hockey teammate used to get his later in the afternoon because he was religious about stopping on the way home from our gams which was usually pretty late at night.
They're banking on (and usually right) that if they give you a small carrot for something free, you'll use the app to buy more stuff. In my experience it really took a lot for me to not buy more than I needed. So I would include the freebie along with whatever else I was buying to try and save a buck or two. Whereas my teammate would get his normal order plus whatever freebie they were offering.
either way, it worked. Many times instead of going somewhere else, I would go there, and use their mobile app and get my freebies.
0. You can set up favorite orders, and get all the nutrition info you care to. And make fiddly custom orders without worrying someone will screw up typing it in.
1. They don't have to pay a cashier to type in your order if you do it for them. And it's more likely to be correct, and more likely to make upsells (would you like to add a drink to your order?). More cynically, you can't complain that their cashier doesn't understand English since you are now the cashier.
2. Price discrimination. Fast food is an unusual mix of fast and cheap, and not everyone values those two equally. Cheap customers can spend time clipping coupons and browsing app deals while fast customers can make their order via drive through without having to futz around while driving. The app replaces newspaper circulars and mailers, esp now that nobody subscribes to local news anymore.
3. You can order ahead, increasing table turn time, spending less time at their limited seating and limited parking. And you spend less time waiting in line to order, waiting for your order to be prepared.
4. It functions as a loyalty program, buy 10 get one free style (but a bit more complicated).
5. Less fighting over coupons at the counter for the franchisees that don't want to accept them - the app just refuses them. And frankly, some of the penny pincher franchisees may not be savvy enough to set up refusals.
6. If a location runs out of something they can flag it and you won't show up disappointed they can't fill your order. Or they can advertise a per-store menu.