For sure it depends on the requirement/use cases, but I think most performance issues have nothing to do with the framework. It's usually N+1 queries, missing indexes or no caching etc... Fix those first.
People chase FastAPI for speed, but if your bottleneck is the database or any network bound work, using async vs sync framework barely matters. Premature optimization is real.
Flask handles production traffic perfectly fine. And by the time it doesn't, you will know exactly where the real bottleneck is.
This is an inevitable stage for every smart appliance: it will either suddenly stop working because it is discontinued, or it will start serving ads. I would argue it the same will happen with smart glasses, autonomous cars, and other smart device
We need sociologists to analyze the people who would actually consider buying such an appliance and tell the rest of us what makes them tick.
Most of the people I know wouldn't let this depraved tech into their homes. Same goes for gas stations where one is forced to watch ads whilst filling up. Most of the ones around me have been dismantled because people have avoided them in favor of stations that had not installed the tech.
By analyzing such people we can teach kids how not to be manipulated by sleazeball corporations like Samsung.
Not a sociologist but I don’t think this one is a big mystery. Off the top of the head I think of two reasons; people who buy things without research based on how a thing looks — you can bet the “demo” in-store isn’t constantly running ads, which should be false advertising — or because they want the thing in spite of the ads to show off to friends, to brag, because their {insert person} got one, etc.
Humans are just jealous moths who lack self-control.
I was watching some cruise crew videos, and it turns out their biggest expense is internet. I'm curious why they don’t just install Starlink to cut costs and maybe sell to guests some "piracy" internet?
---The Starlink dish was secretly installed on the Independence Class vessel’s weather deck, where it was relatively out of view. The network was initially named “Stinky”, but it was later renamed to appear to be a wireless printer – despite there being no such devices aboard.
---The Starlink dish wasn’t discovered until a civilian technician, installing a Starshield satellite communications system, noted the device and reported it to a senior crew member.
Arg, thanks. I literally read that, but there was too much advertising on that site. Maybe the Navy needs to provide Wi-Fi to sailors if it means that much for their morale. And maybe consumer endpoint security needs to be so rock-solid that a sailor can trust their phone/laptop on a voyage.
It's interesting to contrast Starlink on airlines vs Starlink on cruises.
AFAICT, all the airlines rolling out Starlink have made it free on their flights. Which implies cooperation from Starlink -- either Starlink has made "free" a condition of their service, or they've just priced it cheap enough to make free a reasonable option for airlines.
There's no good reason why Starlink for cruise ships should be priced significantly higher than on airlines. So either the cruise lines or Starlink are gouging. Or both. Probably both.
> There's no good reason why Starlink for cruise ships should be priced significantly higher than on airlines.
There's a technical reason for the case of airlines flying over land. Over land, the connection is just up to satellite and down to a ground station.
Over open ocean (whether airborne or on a ship), Starlink has to use their inter-satellite laser hops to eventually get to a ground station. I don't know for sure that Starlink charges more for this mode but if I ran the company I certainly would because those lasers are a limited resource.
The bandwidth of a single starlink terminal is going to be saturated at airplane capacity anyway. The extra number of people on a cruise ship just means service degradation, not excess bandwidth consumption.
The price difference is just based on what the market will bear. Trapped on a cruise for a week, you are much more desperate for Internet. Plus you've paid a lot more for the trip and the fee doesn't feel so large compared to all the other upsells. The cruise often is the vacation, whereas air travel is just the means to an end.
Yeah, but there is a limit to the price the market will bear. At 10k users and assuming a single terminal, and a single price, you are going to not going to be able to price it so as to optimize price vs performance.
Personally I would have at least 2 terminals, a low tier and high tier. I would sell only a limited number of high tier connections, good for the entire trip. Probably included as a perk with first-class cabins. The low tier would be a daily purchase. I mean hotels have done this for ages.
Maybe a dedicated business center with wired (dongle) connection and kiosk PCs, that gets the best bandwidth of all, but you're away from ship activities.
A cruise ship shouldn’t have any issues with having 10 or 20 terminals installed and the clear skies in all directions could mean each has its own bird.
Those 12 terminals still need to talk to satellites. Only a certain number will be in reach, and if the ship is significantly out to sea, those satellites will need to pass data along to others to reach the ground.
We know a densely populated land area can saturate the satellites overhead; it's part of the reason we don't use Starlink in, say, NYC. The same math applies to a thousand cruise ship passengers trying to use it at the same time.
(It will absolutely be much better than the previous state-of-the-art, though.)
Absolutely the density of 12 terminals over what amounts to skyscraper on edge, is high. But cruise ships are not packed into city blocks. The density of terminals over ship-in-ocean sized areas is quite low. The density is very much more sparse than a countryside or farm area. So it feels unlikely to me that you are going to saturate the visible satellites. And those satellites will be very underutilized at their locations not quite over land yet.
Only recently (2025) has Starlink been documented to pass data between satellites and only for new ones. Cruise ships have had starlink for several years so the most common case will likely still be a single hop. Most cruises hug closely enough to the shore (within 100 miles) that a second hop isn't required.
It's not really a thousand passengers, it's just 12 terminals. Sure, those terminals are well oversubscribed vs a normal user terminal, but given the documented terminal-to-ground-station ratio (quite high), I'm pretty doubtful a cruise ship or 2 significantly impacts (or is impacted by) ground bandwidth. The density of users is just at the wrong scale for it to be significant.
I’d guess that concurrent demand would be lower on a plane. A cruise ship has people with nothing to do wandering around, presumably screwing around on the internet being one of them.
> A cruise ship has people with nothing to do wandering around
The major point of a cruise is that there is constantly something to do. It's wall-to-wall entertainment, 24-hours a day.
My 2-year-old was too excited to sleep on a Disney cruise, so we just walked around and found character photos at 10PM. She was too shy to take her picture, so I got my picture taken with her hiding her face on my shoulder.
(FWIW: A cruise is also the kind of vacation that you need to bring some offline entertainment with you. I caught up on reading when I had to stay in the cabin with a sick child.)
Cruises make most of their money now by upselling you on board. Premium drinks and drink cards, more premium restaurants while included buffet and dining room get less love; internet upgrade, excursions, lots of stores selling stuff like diamonds and overpriced fashion, art auctions. It's worse than a floating shopping mall.
Almost, the reason it's free is because competition from t-mobile for the internet provider and that the legacies are providing it free and well almost legacy airlines like southwest and alaska.
Whatever one legacy does, the other do - charge bag fees, the others do too within a quarter. Free internet (Delta afaik was the leader here) the others offer free internet.
Soon it will go the way of having an added fee or being tied to your ticket on the airline, w/ Tmobile its already linked to your phone number.
It seems as it would average out, but I wonder if the equation between "<some %age of> ~2500 people for a 7-14 days" vs. "<some %age of> ~175 people for 2-5 hours" incurs more "costs" for the former?
it's unclear to me why Starlink is free on airlines. I currently pay $30-50 per transoceanic flight for crappy internet. I'd pay 2-3X more for something solid with lower latency.
Regular internet on ships is a flat rate for the ship owner (except for Inmarsat, which is hugely expensive and only used if nothing else works) and the big issue is sharing limited bandwidth with all users. Before Starlink this meant blocking all streaming for our ≈35 people crew, unless you used a VPN, which allowed you to bypass blockage and would get you banned if caught. It's a huge cat and mouse game that burned too much of my time. But then, cruise companies are sleazy as fuck and totally deserve this.
Source: I was a radio operator on Greenpeace ships for nearly 20 years
Starlink can do half a gigabit per beam and point 8 beams at the same cell.
With 5000 people on a ship, that's enough bandwidth for half of them to watch HD video at the same time.
On land they want to spread out the bandwidth as much as possible, but on the ocean most cells are empty or only need a time share fraction of a beam, so they might as well focus on those mini cities.
I don't think a single satellite can saturate a cell, but I think you're right on a couple gateways being able to soak up a couple beams from multiple birds. That might work on rare trans-oceanic cruises. But for the most-common routes, which nose along coastlines, there are other beam uses.
> Except for all the ships that do use Starlink I guess
Hence why they meter and ration its access. Starlink cannot service an entire cruise ship at acceptable speeds for the same reason it can't service a Manhattan city block.
Probably because they've had internet on cruises since long before Starlink existed and it's expensive to change. And why bother if people are willing to pay for it?
Low cost home IC development is something very needed for agriculture. If we think about current and future farming equipment, it's digital. We need to provide them the ability to repair themselves and mod.
You can just get finished Microcontrollers that you can program yourself for a fraction of a fraction of the price to make them yourself, and orders of magnitude more capable. You will not be able to make a chip more capable than an ESP32 for less than $2, so how would making an IC yourself help you?
I imagine it would be in preparedness for a time when you could no longer get such powerful chips so cheaply and quickly, or for one where you no longer trusted the chips you could get for some reason.
How about we let them flash the ICs that they have first? Or allow them to change the maximum speed on the vehicle without having to go to the service center and paying 300 to 500$.
Why are we talking about low cost at home IC development for farmers while we don't let them do even that.
Forcing companies to open source their software is not possible, but making sure we can replace each component after warranty? There are strong right to repair movements, it's just matter of time.
But you still have to reverse engineer the IC before you can replace it - and once you have that it's still cheaper to have it manufactured in an existing Fab than to build your own.
Forcing companies to open source their software is certainly possible for some senses of the word "possible". You could go a long way in that direction just by changing the copyright law to not cover software, or to only cover software if the source code is deposited with the Library of Congress. Or you could change product liability law to declare products shipped without complete, compilable source code for their firmware to be "defective". Right now the political support for such changes isn't there, but that can change over time.
More likely we're going to go in the opposite direction, though.
My father grew up on a farm and I wholeheartedly agree. Unfortunately, this is a step in the right direction but the goal is still a long ways off. Farmers don't have a spare $50k sitting around to build hobbyist IC fabs in the barn.
Any IC that would be practical to DIY is available for <$1 and you could probably get something 1000x more powerful for nearly the same price. Making chips isn't the issue here.
reply