Nit: Most food delivery workers in NYC are on e-bikes or mopeds. There are surely some drivers but it is prohibitively difficult to use a car for food deliveries in nearly all of Manhattan and decent portions of Brooklyn/Queens/the Bronx.
As it is there is public outcry about dangerous and aggressive riding behavior from food delivery workers. It is not strictly a bad idea to tilt the incentive structure so that while you can earn a little extra on tips for completing more deliveries, it's not such a huge part of your earnings that you need to cut corners, ride the wrong way on one-way streets, ride on sidewalks etc. to hyperoptimize your delivery rate.
I stopped using door dash when I watched multiple delivery drivers go pick up my food, sit and wait for a bit, drive in the wrong direction, pick up food from another restaurant, drive in the wrong direction to another location, then deliver my cold food.
At the end of the day this is all driven by market dynamics that affect everyone. Uber Eats doesn't have a secret sauce that makes it cheaper for drivers to deliver food in cities with lower restaurant density. Someone has to pay for gas, someone has to pay for drivers' time. Even in large cities Uber Eats is now more expensive than ordering directly from restaurants in the past, why would it be different for cities with fewer restaurants?
Oh I don't think anyone said they're making it cheaper. But a lot of restaurants now have expensive delivery instead of no delivery. Before, there were very few places where most restaurants would deliver.
You are describing a completely different service though. Prep-meal delivery also works, but it's addressing a different usage pattern.
In the same way you could propose to replace Uber with... buses? 15 people get on the same bus at a specific time and get dropped off within 5 miles, thus optimizing the process.
I've seen the Uber drivers come in. They always grab like 4 or 5 orders, and possibly drive to a 2nd or 3rd restaraunt before they start delivering.
Its not like those UberEats drivers have a big penalty if they arrive late or if the food is cold.
-----------
What I'm saying is, consolidated food delivery services is *already* what we have with DoorDash/UberEats/whatever, they just lie to you about the details.
The reason why Foodsby works (albeit on a smaller scale) is because they're honest about it. There's nothing wrong with consolidating orders to minimize driving time, the problem occurs because those other services _PRETEND_ they're a 1-to-1 order service with dedicated drivers. IE: Its the lies where things have gone wrong, not necessarily the practice.
If all the drivers are double-booking / consolidating orders anyway, then work it into the model. Embrace it, rather than pretend otherwise.
>I've seen the Uber drivers come in. They always grab like 4 or 5 orders [from the same restaurant], and possibly drive to a 2nd or 3rd restaurant before they start delivering.
Having done Doordash/Ubereats in the past myself that is ABSOUTELY not the norm, god I wish it was that streamlined. Picking up a single order and delivering it straight to the customer is by far and away the most common scenario for the drivers.
They do have stacked orders which are the multiple pickups you're talking about, but I've never had more than 3 orders "stacked" together, and I would say it's more common to have stacked orders from multiple different restaurants rather than multiple orders from the same restaurant. And from the driver point of view stacks suck because almost always only one order in the stack will have a tip, the others will be no-tip orders they couldn't get someone else to deliver by themselves.
Doordash also recently changed how they pay out on these stacked orders to the drivers detriment. It used to be you'd get the base rate for each order in the stack, so 3 orders stacked together would be the base rate x 3 + whatever tips by each customer, but now they pay them out as one big order no matter how many orders are stacked together, so you get a single count of the base rate even if you're delivering two or more orders in a stack, which again are usually only a single order with a tip, so you're effectively delivering the other orders for free.
Thanks for the anecdotes. Good to hear how things work from someone who has actually experienced it.
All I've seen are the big orders that get handed to someone who immediately runs out the door and into a car that they left running (with the keys in and everything), lol. They're obviously stressed and trying to make time.
I guess the big orders / stacks are more noticeable to me and obvious what's going on. If a more "normal" driver comes in, it looks like any other internet order (like my own, except I'm picking it up personally).
You are partially correct describing that some delivery drivers already batch their orders. But that's still a different use case. Foodsby can't scale to 15 min increments for all restaurants in the city. While some people can preplan and are OK ordering food from a specific restaurant that Foodsby works with, most people don't want that. How many people want to order food from an expensive Spanish (i.e. Spanish, not Mexican) restaurant I order from sometimes? So my options are going to be: #1 not ordering from this restaurant because it's not a common food preference and is expensive or #2 wait for 15 people to join the order for this restaurant and get food hours or days later.
It's a good business idea, it's just a different usage pattern. I order food when I am hungry. I don't preplan, don't like food from the majority of popular local places (pizza, Mexican, Chinese).
> Foodsby can't scale to 15 min increments for all restaurants in the city.
For some definition of increments and "all", yes they can.
Restaraunt#1 delivers at 10:15am.
Restaraunt#2 delivers at 10:30am.
Etc. etc. etc. Covering the entirety of lunch hours. This is how it works in practice, today.
If you're in a location with lots of Foodsby usage, then you might have 3 or 4 different Foodsby locations to check within a reasonable distance, which dramatically increases the restaurants and timeslots available. Like 1x Foodsby is already fine, but if you're in an area with 5x walking-distance Foodsby dropoff points like I am, things start to get really convenient, and the selection becomes dramatically wider.
It also means for the drivers, that one "trip" can hit 3, 4, 5 offices in one drive. I'm sure that on Foodsby days, these drivers are delivering multiple dozens of meals. In fact, the *MANAGER* of one local Restaurant was the driver for one of my recent orders (we recognize each other's faces because I visited his restaraunt a lot, so it surprised me to see the manager making delivery runs). So its more fulfilling work than typical grunt labor, since they're making so many deliveries on relatively low effort. If he's got ~30 orders, that's $60 ($2 per meal) in less than 30 minutes of driving/delivering, which is certainly more money flowing than most UberEats / DoorDash setups.
--------
Now yes, you may be arguing that "its not what you want". But... when that Spanish Restaraunt says "We're offering $2 delivery (no tip) 3 days from now at X-oclock"... I think you'll be thinking of using Foodsby that day.
Or maybe you check the website to see today's Restaraunts and whether or not your favorite is on the list.
-----------
In any case, _THIS_ is innovation. Actually playing with models and finding things that are better for everyone (chefs, restaraunt owners, drivers, users) as a whole. I'm sure other models can work too, for some sliding scale of individualism, bulk deliveries and whatnot.
But as far as the personalized 1-to-1 service? Its dead, its so dead I'm convinced it never even existed. UberEats _never_ promised a driver on the standby ready to personally serve you, and months/years of using the service has made it obvious to everyone.
There's only so many times that I get a meal 1.5 hours too late that causes me to give up on UberEats (and similar) services.
Interestingly in Silicon Valley, there's an angel funded startup founded by an ex-Google foodie who worked with some of the more notable Asian restaurants (the super popular ones with long lines).
She arrange it such that users from a given neighborhood could order via the website app by 4:30 from a specific set of 2-3 restaurants per day. The orders from a given neighborhood are bulk ordered for 3-4 drivers to pick up from the 2-3 restaurants and delivered to the same neighborhood between 6-7pm.
The list of 2-3 restaurants for a given neighborhood are shared a week in advance and in cooperation with the restaurants so they can handle the surge. Since the restaurants no it will likely be reheated the packaging is optimized.
Because there is a rotation of the restaurant - it paradoxically avoids the tyranny of choice issues with picking a restaurant and actually feels fresher in a discover new restaurants type of way.
During ZIRP/Covid the menu prices were sometimes lower and there was no delivery fee. Post Covid they do have a membership option or a very modest delivery fee.
Hello, what is the name of this? I am currently working with a few restaurants to launch almost the exact model. The idea is to transition from on-demand to in-advance.
Uber was experimenting with that exact process before covid. For a lower fare, you let Uber dynamically re-route your driver to pick up more people headed in roughly the same direction as you. Worked alright, maybe took 1.5x as long for you to reach your destination while you paid about half as much as you did usually.
Busses work, though. If you removed busses, there is no amount of Ubers you could add to the roads to "fix" the system. It's clearly impractical for each office worker to pay for even half an hour of someone's time every day to deliver a meal just to them. While most counter services can serve 20 meals with the same half hour of wages.
Uber Eats is doing it here in London with "ghost kitchens". 15 fast casual brands cooking (mostly reheating really) from 3 kitchens in 1 building. A much better chance for delivery drivers to be able to pick up multiple orders. It still isn't enough optimising though, as the destinations are still scattered.
Do you know by any chance if wiim is capable of streaming music independently, without relying on a separate computer? I.e. my smart Denon speaker can stream Spotify, so if I start playing on my laptop through that speaker and then close the laptop, streaming continues. I don't have to keep my laptop running. This doesn't work for Apple Music and I am looking for a device that can stream Apple Music without an additional computer.
Unifi is just their name for this series of products that all work together with their (free) controller. So you buy an AP, a switch, a router, all from the Unifi line of products and they all get setup from a single controller and work together pretty effortlessly
I see, thanks. So basically a suite of networking hardware solutions? What I found the most confusing is that they sort of differentiate UniFi from WiFi. E.g. "Instant WiFi for retail POS" - yes, like any other WiFi router? Or is this box going to connect to a global mesh that's called UniFi?..
For the "Instant WIFI for retail POS" it could be describing a meshed wifi solution, where it then acts as a wireless backhaul and bridges to the LAN port you can connect to the POS system that doesn't have wifi built in.
Exactly, I only understood that it's basically a WiFi router by reading this thread. I guess I am not their target buyer... still, would appreciate more clarity.
Read this several times. Still don't know what I am looking at. What is "UniFi Networking stack", what is "UniFi Network"? Is it a WiFi router? I use a TP-Link router at home, and they don't sell it as "TP-Link Networking stack".
While TP-Link offers simple routers and basic networking equipment (which I’ve found to be 100% reliable at home), they do also offer their Omada based products which seem to compete with Ubiquity’s products.
All their Omada products can communicate with a controller which can auto configure the devices and actively coordinate handover of clients between WiFi APs.
I’m running it in my house and I’m pretty happy. Through the controller web page UI you centrally define your VLANs and wireless networks and then it updates all your equipment configuration for you.
I don’t think I’d ever bother with their gateway product as pfSense seems to be way more capable.
More complex networks (e.g. in commercial settings, or the homes of I.T. nerds) will often consist of a router plus one or more switches and/or wireless access points, probably spread out across a site with multiple ethernet drops. A typical router sold for home consumers (like yours I'm guessing) is a combination of all three of those in one convenient package, which is enough for most home use cases.
Companies like Ubiquiti (under the UniFi brand), Meraki, etc. make these products such that they can all work together as an ecosystem, e.g. so you can log into a single dashboard and manage the network as well as every individual device's configuration from one place. This is the difference between a so-called "managed" switch (or wireless access point) versus an ordinary dumb one. UniFi also makes PoE security cameras that are managed through their ecosystem in the same manner.
This sort of ecosystem is useful for people doing I.T. in commercial settings. You can use a single interface to manage a network in a huge office building with hundreds of devices, or to manage lots of smaller networks spread across different sites. This "UniFi Express" product seems more suitable for the latter, e.g. in cafes or small retail settings where you might just use it on its own or add a small number of additional switches/APs. It's similar to your home router+AP combo, but it also contains the management software I described before which is capable of adopting more UniFi devices and provides remote administration.
I am not much of a networking guy but decided to try the Unifi stuff several years back after some frustration with one of the consumer mesh wifi things. I found the Unifi stuff incredibly confusing when reading about it prior to buying some. Some friends at work who are more networking savvy were very keen on Unifi were very helpful though, and it turns out not to be as strange as I thought, although coming from the garden variety consumer wifi router boxes, it seemed bizarre at the time. I'll try to give an explanation along the lines of what they gave me.
In a wifi router like your TP-Link, the control plane software is running on the box with the the switching hardware and wifi ap, so you've got a little single board computer running web server for the UI, and all the random dhcp/dns/etc and other doodads that can run on them.
In the Unifi world, you've got all the same functions, routing/switching/wifi etc, but instead of sharing one box, the functions are spread across a number of different devices.
As with a combination router/wifi/switch, the important part to a user like you or me is that control plane software- you plug the thing in, point your browser to 192.168.1.1 or whatever, and set things up. The Unifi world has this too, but that software component doesn't need to run in any specific place in your network. So for example, you could buy two Unifi Access Points which do nothing but talk to wifi clients, then you would need some kind of device capable of going your routing, and you might need a switch as well.
Ubiquiti sells a variety of little routers and switches that can perform those network functions, but which don't have any compute or storage resources that would be necessary to run the control plane software. However, they also sell little gizmos like the Cloud Key which can run the control plane software- it's just a tiny server with some flash storage and an ethernet port. I'll refer to that thing as The Controller.
When you change some settings on your TP-Link, the web UI app is twiddling with the the routing/switching/wifi/etc hardware or software on the device. In the Unifi world there's a web ui as well, running on the controller, but when you change a setting in the web ui, the controller decides which devices need to have their configuration updated, and sends out new configuration to them over the network.
Here's where I think things get confusing. The Controller software package can run on a wide variety of devices which are so different that the whole thing will seem nonsensical if you're used to regular wireless routers. You can buy a CloudKey and connect it to your network. You can download a copy of the Controller that will run on a Linux box on your network. Or you could do the same thing but have the Controller running on a machine that isn't on your network at all, like an EC2 vm. Or, Ubiquiti also sell some devices which combine two or more functions into a single device, like some of the "UDM" family of devices have compute and storage resources in addition to the switching/routing/wifi hardware, and have The Controller software installed in advance.
To give you an example of how flexible the controller placement is, I have a little Synology NAS that is able to run Docker, and on it I have an image that contains the Unifi Controller, so when I go the web ui for my network, I'm talking to a containerized web server on the NAS, which is managing the configuration of my devices, which are a router (I just replaced my USG 3P with a UXG-Lite yesterday), a couple of their little inexpensive switches, and a pair of Wireless Access Points.
What I like about this model is that I'm able to update pieces of it as I need to, and usually the individual pieces are fairly inexpensive. But what I dislike about it, as some other respondents on this thread have complained, choosing which devices you need is confusing as hell. They sell at least one machine which has a wifi AP, switch, router, and controller all in one box. Why not get that? The reason, I believe, is that many of the people who use this Unifi stuff are managing a bunch of networks at a bunch of different sites, like maybe at a bunch of retail locations or restaurants, where its way more convenient to have the controller running offsite, but then they decide to install some stuff at home and need a Controller which needs fewer resources so a Cloud Key or just running the stuff on your desktop would be ok.
This flexibility means that there's no single right set of hardware, no single best product, etc. Especially if you don't need multiple APs, I think a single-box wifi router will provide equal or superior performance with much less trouble, but once you need multiple APs, the Unifi stuff can be compelling if you're comfortable with the architecture, but I think it's difficult to decide which hardware bits to choose and what the best place to run the controller will be.
anyway apologies for the length- I found all this very confusing initially as well although I've grown fond of the Unifi stuff and thought it might be worth writing the whole thing out in case its useful to somebody considering this stuff.
There is a distinction to be made between a parseable string and a valid URL in more complete terms. While it's nice to have a new standard function that checks whether a string is parseable as a URL, it can't, of course, validate that string in terms of whether the TLD is a real one, and not something like "https://super.hacker/".
So maybe the post should have called it what it is: validating parse-ability, since the method itself is called "canParse()".
I had success in the past hampering the process by linking to various legislations (e.g. GDPR, CCPA) and asking my PM to reach out to legal to approve the implementation. At larger companies legal is usually pretty cautious and takes a very long time to respond.
My guess is that the author chose to approach this in a softer way, without shaming the designers. He is leaning into education without assuming bad intent.
Not going to lie, this might be more effective than shaming designers.
