this post was submitted on 12 Nov 2026
879 points (99.7% liked)
Technology
89678 readers
4224 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 3 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
There are good reasons to use bits for engineers, but residential internet consumers are not engineers. In no other context do you market incomprehensible engineering details, just here because higher number is better marketing. Almost everyone I know is confused why the speeds they pay for don't match the speed of downloads in Chrome. Marketing in bits is anti-consumer, since average consumer does not understand what actual download speed they will get when buying.
It’s not anti-consumer to advertise a product’s capabilities in the most accurate way possible. When you pay for internet bandwidth you’re not paying for download speed, you’re paying for a transmission rate. Advertising internet plans based on the speeds you get in Chrome would not be possible and therefore any attempt to do so would be deceptive to the customer.
The Internet bandwidth they pay for will never match the speed of downloads in Chrome, even if you convert bits to bytes.
What you’re missing here is that you want the number you pay for from your ISP and the number you can measure downloading a file at to match. But they never will, because they’re two completely different concepts. One of those number is a constituent part of the other number. They’re not interchangeable.
Why wouldn't it be possible? Convert Mbit/s to a reasonable real-world mb/s estimation for download/upload to a reasonably close location, with some losses built into the estimate.
You’re asking why it’s not possible, and then explained a work around that would be needed because it’s not possible.
What you’re suggesting is that instead of selling a service to you based on a quantifiable number based on things within their control that you can reasonably hold them accountable for, they should sell their service to you based on a series of educated guesses about what you could potentially get based on a number of factors entirely outside of their control.
Elsewhere in the thread it looks like some ISPs do exactly that! Not sure why you are against the mb/s estimate.
Edit: thanks for the detail re: Mbit in your other answer!
I still think an estimate of Mbit/8 e.g. mb/s would still be a more useful thing for consumers to compare to the download rates they're actually getting.
The first speedtest I ever used was from the ISP and gave results in Mbit -- which confused 12 y.o. me because I was not getting anything close to that when I downloaded files.
If the speedtest said 1 mb/s and then I got 0.8 mb/s downloading from a server, I wouldn't call ISP support, but I think we did call support when the speedtest said we were getting "10 M" but our downloads were closer to 1.1 mb/s.
Listing the bandwidth in bytes would be less useful because they still wouldn’t match the download rates they’re getting, but also your internet bandwidth would now be in a different format from every other networking device in your house. Paying for 125 Megabytes but your downloads only ever being 100-112MB/s is still going to prompt those service desk calls, and now going to the store and trying to buy a router would be more confusing as well because you’d want to buy one that matches the speed you’re paying for but now your bandwidth is in bytes but all the routers measure their throughput in bits and you’re right back to being confused.
If you’re streaming video from or to the internet and run into issues it’ll be harder to figure out what the problem is because your ISP bandwidth is in bytes but your video bitrate is in bits.
Changing one factor in your internet solution to bytes while every single other component involved measures it in bits is just not by any stretch of the imagination going to result in less confusion overall. You’re robbing Peter to pay Paul, but Paul is still being shorted.
At the upper end of that range, you are only getting 90% of their advertised bandwidth, even in ideal conditions. Is that actually an acceptable loss? Maybe they shouldn't advertise 125 mb/s if they can only deliver 90% of that in the best case.
Routers, network cards, and Ethernet cables being "gigabit" or similar is also confusing, but i see your point there.
I still don't see the harm in providing both formats, thought i do see the point in not switching to mb/s only.
In the example you posted, it would be 1000Mbit/s down, which translates to a real world speed of 110-112 mb/s .
They aren’t delivering 90% to you, they’re delivering 100%. You are getting all 125MB/s, the full amount they advertised to you. The fact that you’re confused here is exactly why they do not advertise bandwidth this way.
No, it’s not confusing. Those devices pass bits between each other. It would be confusing if you measured the bits they’re passing in Bytes for some reason.
The real world speed is 1000Mb/s down.
It's not some reason, the reason is that file transfer uses bytes.
Hiding it is anti-consumer.
What good is it that i can reach the local node in 1000Mbit/s if their first hop to an internet server slows the connection to 90%?
If i upgrade to the 2500Mbit/s connection, am i only getting 90% of that too?
They should advertise what internet service they're really providing.
The job of computer networks and network devices is not file transfer. Its to pass information, only a subset of which is for file transfer. Networking devices pass pass bits between each other, so it only makes sense to measure how many of those bits they can pass per second.
Hiding it is not anti-consumer. Nothing is being hidden. The capabilities of the devices are simply being accurately described based on the job they perform and the way they perform it.
The first hop to an internet server doesn’t slow the connection by 90%. The connection remains at 100% the entire time.
If it is 100% why is it 90%?
You’re referring to two different things. You are using 100% of the bandwidth they’ve sold to you, which is 1000Mb/s or 125MB/s. Thats the “speed” at which you’re transferring data across the network.
90% of that bandwidth — that you are using 100% of —consists of the file data you’re downloading.
This is why those two numbers are not interchangeable, and why simply converting to bytes won’t result in the numbers you expect as a consumer. Because of the very nature of how computer networking works.
If the file is downloading at 112 megabytes per second, how is the link maxing out at 125 megabytes per second?
That is a very good question.
The answer can be simplified down to what is really two answers.
The first is: Overhead. When you download a file over the Internet, it’s broken down into many many little messages called packets, which you can think of like postal envelopes. Each envelope holds pieces of your file, but that’s not all they hold. They have to hold the senders address, the receivers address, your zip code, your town, your state, your country, etc. all the info it needs so the postal service (Networking equipment) knows how to deliver it. Let’s say the file you’re downloading is an image. The person sending it to you takes a piece of that image and puts it in an envelope and mails it. They intend to do the same with a million other pieces of the photo until you on the other end can take all the pieces together and look at the photo. This postal service of sorts will also put that envelope inside other bigger envelopes to help it move through the system, among other things. These encasing envelopes get put on and stripped off at various points during the journey, but each of those envelopes are tangible objects. Just because the envelopes themselves are not part of the image you’re sending/receiving, doesn’t mean they don’t take up space in the back of the mail truck. Also before they even started sending the pieces of the image, you and the person on the other end had to exchange a few letters agreeing to the image delivery. You also frequently send letters to each other to keep track of what was sent and what was received, etc to make sure everyone’s on the same page every step of the way, and not all of them contain pieces of the image.
To make things more complicated, “Downloading a file” can mean a great many things to a computer. In the case of the photo, what happens if some of the envelopes get lost? Luckily the postal service has a mechanism that helps the recipient notice that one of the pieces of the image you were expecting never arrived, so you can send a letter back informing the sender so that they can send that piece again. After all you don’t want holes in your picture. Let’s say that works — you now received that piece. Well that piece took up space in the postal truck the first time it was sent to you, and it took up space again the second time they sent it to you. There are other ways of downloading a file though where you and the sender agree not to resend any lost envelopes, you’ll just deal with the holes in the photo. This obviously takes up less space overall because there are no duplicate envelopes being sent, and this decision is made without consulting the postal service. It’s strictly a decision made between you and the sender.
The second answer, which we will keep the letter metaphor for, is that not every piece of mail you send out or receive relates to that image you’re sending. You get a bill from your doctor in the mail, Your birthday is tomorrow so you receive a letter from your grandma. All kinds of stuff. The postal service is delivering all of it to you and it all takes up space in the truck a long side the pieces of that image.
Now you paid for a finite amount of space in the mail truck, and during this time you’re filling that space you paid for down to every last inch. You filled it with the pieces of the image in the envelopes, the envelopes themselves, the envelopes that don’t have pieces of the image in them but are related to the image, letters and the envelopes that contain them that were mailed to you by your doctor and your grandma, etc.
You filled 100% of the space you paid for in the mail truck, but only 90% of that space is taken up by the pieces of the image. The other 10% is filled with duplicates to earlier lost pieces of mail, the envelopes themselves, and the unrelated letters you’re receiving from other people.
This is why the postal service advertises to you and bills you based on space in the truck, a known and quantifiable amount of space they can commit to you, and not by how fast that particular image or anything else will be delivered. The postal service doesn’t exist for you to send images, it exists to send ... stuff. All the stuff, it doesn’t care what. They give you the space, you decide what to fill it with. How much of that space is taken up by just your image pieces is dependent on the image your sending, the way that you and the sender agreed to send it, how full the image piece envelopes are with other sets of instructions meant for you and the sender that doesn’t involve the postal service at all that you stuff in next to the image piece, etc. That’s your prerogative. What the post office is concerned about, and should be concerned about, is that the space you paid for is available to you on the truck, and that the truck will successfully deliver your mail from your local post office to the next post office on the route.
You can give an accurate enough estimate. Yes, you may have to make some assumptions on protocol and server. And yes, you will be a few percent off. But so what? It is much more informative to the average consumer than transmission rate. In the first place, transmission rate is also a meaningless metric without information on overprovisioning, latencies, etc. Except instead of an understandable but slightly inaccurate download speed estimate, you have one engineering parameter that is unhelpful to an engineer anyway without more information as well as not understandable to the average consumer.
PS: I am not saying I can't include engineering parameters, but it shouldn't be the main marketing number.
Accurate enough for who? It wouldn’t be accurate enough for me. And it certainly wouldn’t be accurate enough for the consumer who you described as expecting that estimate to match reality. And worst of all, it wouldn’t be enough for any consumer protection agencies or remedies. So many assumptions would have to be made based on factors entirely outside of their control in order to sell to you based on your file download speed. And if you allow them to sell you a product based on assumptions of things outside of their control, then you then have no recourse to hold them accountable. I don’t want companies to sell me a product based on a number offered via a shrug. I want a promised number I can measure against what I’m getting.
Transmission rate is the only metric with any meaning in this context, because the transmission rate is the product they’re selling to you. The overprovisioning and etc is actually accounted for in the transmission rate they’re selling you. Thats why they almost always overprovision your connection based on what you’re paying for.
It’s as if you’re asking for a car dealer to sell you vehicles based on how many how many people you can move between two locations per hour, instead of the number of seats it has and the speed it can go. They would have to start making guesses based on the average waistline of people in your area, the speed limits of the roads in your town, your driving ability, what routes you like to take, etc. that might be sorta handy because you don’t have to do that math yourself now, but it’s not going to be all that accurate to you specifically, and if that number of people is lower than what they estimated, it’s kinda on you because they don’t know if your friends are petite or 600 lbs, etc etc. But if they sell you a car that seats 5 people of normal healthy size and it can go 100mph, and they deliver you something that seats 4 with a top speed of 80, they’re accountable for that.
What are you talking about? Overprovisioning literally means selling more transmission rate than what the upstream lines can accommodate. It is not accounted for. Therefore, you only get your advertised transmission rate if your neighbors are not fully utilizing theirs.
How can it not be accurate enough when it is just a conversion from transmission rate via a formula that converts to bytes and accounts for expected protocol overheads? It's the same number, converted to easily understandable units. If it is not accurate enough, neither is transmission rate.
I’m talking about overprovisioning, where your ISP actually provisions you for more bandwidth than you’re paying for. It is accounted for.
You’re talking over subscription.
Are you asking me how a flat base number derived direct from the source is more accurate than a different number that takes that first number and then adds a bunch of guesswork on top based on a bunch of unknown variables? Well that’s because in your head “expected protocol overheads” must be a simpler and more quantifiable number than it is in real life. For starters, which protocol? Downloads happen over TCP and UDP, both have completely different overhead profiles. What about the protocols above that? How are we proposing the file is being downloaded? Tor? SMB? NFS? HTTP? MLS stream? Which OS is the client? Different operating systems have different TCP window scaling logic. Different devices have different MTU sizes. The Internet service you’re buying can be used for all of them. Are we picking picking one set based on the most common use case? Averaging them all together? Weighting them based on common usage? Uhhh or we could just advertise the product based on the product the ISP is actually selling you, the transmission rate.
Your entire argument here is like being upset at McDonald’s for giving you the nutrient values of their food, like calorie counts and detailed macronutrient breakdowns, because you think instead they should just tell you how full each item will make you.
I really can’t break this down any simpler for you than I already have.
Dude, what is your fixation on the protocol parameters? I said average. Yes, for the purpose of getting the exact speed Chrome downloads shit, it will be off by a few percent. Who cares? The weight of the food you buy is allowed to be off by a few percent. The calories in food are already an approximation based on energy released while burning, not digestion. Every measurement in the world has error bars.
But a number most people can understand is a lot more useful than a slightly more accurate representation of the same number most people don't understand. I am sure you can still find the details in a spec sheet if you feel too smart to use the simple representation and too lazy to convert it back.
Ok, let's say I used the wrong term for the sake of the argument. Point, is (most) ISPs don't guarantee your full ratelimit will be available to you. Which you clearly understood yet did not address.
I mean, wasn't this discussion started by whether bits per second or bytes per second is deceptive or misleading in order to make people think "woah, bigger number means better internet" ?
Also, how is a division or multiplication by a factor of 8 the difference between what most people do or don't understand? It's a primary school mathematical operation. If people cannot understand that, how are they going to understand that different downloads or uploads inside different applications provide them with different (byte-based) values depending on protocol and server performance of whatever service they are using?
It really is the truest and simplest unit to use that the ISP can reasonably guarantee to you, because on the lowest level, whether using fibre or copper, we send bits and not bytes across the wire.
No one who was not specifically taught this will notice the difference between Mbps and MBps since it's the same fucking letters. Let alone know the difference is division by 8. Not everyone has formal IT education and bits never come out in daily life of most people except for internet speeds. You are expert in this field. You massively overestimate how much average people know if you think they recognize the difference and know how to convert it.
Why do you keep insisting they need to? An average person will not care much about the difference between a download taking 6 minutes or 7 minutes. They will care about the difference between 6 minutes and an hour, which is about the difference if they misinterpret bits as bytes.