*Actual user storage less - The USB drive capacity cheat

[@maverator]

Moral of the story: Never trust anything on a package that has an asterisk next to it.

[@matthewhumcke3182]

well said

[@_lun4r_]

Why pinned Unpin this comment

[@ZENITH_System_3]

Very true*

[@EeekItsSnek]

​@_lun4r_ ?? What the issue??

[@ivok9846]

moral of the story: Mac recognizes usb flash within one second. that's fast.

[@an2qzavok]

I like how "actual user storage less" doesn't actually specify how much space you would be missing. Buy 256 Gb drive, get only 2 GB, "you've been warned, it's right there on the packaging, no refunds"

[@KRAFTWERK2K6]

This is basically how it feels most of the time... This is why i think there should be a standard that guarantees the user to get at least 98% of the promised capacity. They will use this "decimal" bullshit excuse for all times if nothing is done about it in legal terms.

[@rich1051414]

@@KRAFTWERK2K6 And then they will say 97% rounds up to 98%. And then 95% basically rounds up to 97%, which is already known to round up to 98%. No. It needs to be at least 100% of stated capacity. No compromise.

[@thepwrtank18]

@@rich1051414 between -100% and 100%, 0% rounds up to 100%

[@cs8712]

you'll own nothing, and you'll be happy [to write data on it]

[@bjarnenilsson80]

Well if you buy a 256 Gb ( lowercase b means bit not byte) you would exoect it to show up as about 64GB but I'm nitpicing and probably missing your point, oh well life gies one

[@ryuquen]

I think those bigger USB thumb drives are hiding a more hideous problem by "cheating" on the available space. Bigger thumb drives tend to use TLC even QLC flash chips, which have much shorter write endurance than smaller (8G or smaller) MLC chips. So these bigger ones (including SSDs) have to do bigger provisions for their internal controllers for detecting and "masking" (moving data from failing blocks to better blocks) any defects.

[@claudiodiaz9752]

Then they should make a 132gb drive, and make 128 available.

[@gabotron94]

I was looking for a spare flash drive last week, and noticed everything was over 8GB and also super cheap. I felt concerned... they had to be hiding a compromise somewhere. And now I know what it was!

[@kaitlyn__L]

@@claudiodiaz9752 that's basically what they used to do. I would be surprised if the 2.01GB reported one didn't have 16MB, 32MB, or 64MB set aside (it was still necessary, just not to quite the same magnitude as newer denser cells). The new status quo is purely penny-pinching and profit-maximisation.

[@eDoc2020]

@@claudiodiaz9752 That's just not economically feasible because the dies are all powers of two. You'd need an extra four gigabyte chip which will add at least an extra 50% to the chip count.

[@eDoc2020]

@@kaitlyn__L I thought they usually relied on the discrepancy between binary and decimal units to reserve enough for basic wear leveling.

[@just.oblivious]

In Disk Utility you were mostly looking at the size of the first partition, not the actual media size. There might be some unpartitioned space left on these drives that could easily be reclaimed by repartitioning them (though I expect them to max out the primary partition right from the factory).

[@prodbyfaith]

MacOS adds 300 MB EFI partitions to every single drive it comes in contact with, which you can't see in Disk Utility, so yeah. (By "in contact" I mean format to GPT)

[@pauledwards2817]

They should have been erased first with the actual device selected not the partitions. As someone who uses dd for block read and write size differences would be quite noticeable and a restore would not be possible. What you are examining is not quite correct. Perhaps a more annoying issue is the decimal versus binary measures. This usually means dd images from spinning disks cannot be written to ssd for example.

[@HappyBeezerStudios]

@@pauledwards2817 I would go further. Check the partitions that come on it, then reformatting it to one's use, and checking again.

[@fungo6631]

The discrepancy also could be the amount of provisioning for wear leveling. Some storage is traded for better robustness against bad sectors.

[@james-5560]

If they decide to do that though they should advertise it at 6GB lower (for example) or add it on so it matches the description it is sold with.

[@belg4mit]

Not "could be" that's literally what the second email said. The issue, of course, is whether or not that's necessary or useful on removable storage from the consumers perspective.

[@cydragon2.099]

could be

[@Stealth86651]

Was going to say, 124 usable, with 4 as "backups" in case memory modules fail. Is that sort of what you're talking about? I actually used to wonder if that was implemented in the more robust versions/products, seemed like a good option if you can't create or afford to create "perfect" memory modules. In that case they could just label it as 124/4, with the first number being usable storage with the second number being backup/dedicated modules for speed or redundancy.

[@Motolav]

It's likely excessive over provisioning so they have less devices to RMA

[@feieralarm]

With drives larger than 32GB, I'd actually see firmware implemented over-provisioning as a benefit, but they do need to make it clear on the packaging. I'm also in no way surprised that smaller drives and no-name brands just don't bother with over-provisioning. As if they care that your $1 4GB dies after a year.

[@eDoc2020]

All of these remotely modern drives have some over-provisioning. The 4 gig drive will present (just over) 4,000,000,000 bytes even though the raw chip will have 4,294,967,296 bytes.

[@vdochev]

It is possible that they also calculate the drive's cache and buffer size which are not usable storage space. Or they just had (for example) defective 256 GB chips and they cut them down with a different firmware to 128 GB or less. But I agree with you that they should market them with their true usable storage size, regardless if it's standard or not. EDIT: These ultra small drives that you say hate, have a good application for me. They are great for listening to music in your car because they don't stick and I don't think just reading from them small files will make them that hot. In any other application I prefer slightly larger drives.

[@dashcamandy2242]

I can verify firsthand the Sandisk Cruzer Fit series is a great USB drive for car stereos. They sit very close to the faceplate, making them less-susceptible to vibration disconnects and flailing hands... They come in capacities large enough to put almost my entire music collection on with room to spare (and I have a LOT of music ripped from my CD collection, cassette transfers, etc.).

[@HappyBeezerStudios]

Using bigger chips with some bad cells on smaller sticks is not uncommon. Gives them a way to use those "bad" chips.

[@miguelhamrol6567]

They should put a bit more storage than the advertised one, to account for leveling and moving data around. That way you would get the advertised capacity or more as available to the user. I remember back in the days of VHS cassettes, you would always get one or two extra minutes from the advertised running time.

[@butterh2]

better margins for these corporations if they can not give you the full capacity and say user storage less instead of giving you more for wear leveling purposes

[@ajddavid452]

they already do that, it's called overprovisioning

[@CompComp]

I wonder if this is why some companies don't advertise in the standard binary capacities (2, 4, 8, 16, etc.) I've seen a lot of drives over the years sold as 30, 60, 120, 250, 480, 960gb.

[@CompComp]

@gorak9000 Nope, definitely SSDs were sold in the rounded down capacities I mentioned. Look up Kingston A400 and Samsung 870 evo.

[@mparagames]

@@gorak9000 yeah no, SSDs do that too, I have one that is advertised as 240GB

[@neoqueto]

SanDisk's explanation does hold some water. While the full 128 gigs is not available to you as the user, it is available to the drive's firmware. And the user benefits from this. It's called "over-provisioning". As bits are written to the NAND cells, they degrade over time. To alleviate some of the damage due to those write cycles, Flash storage manufacturers define some space on the drive that contains basically spare cells. In an ideal world, it would be "extra" space, ensuring that the user has the full sticker amount of gigabytes. Or letting the user control over provisioning space. Some SSDs (like Samsung's and Crucial's) allow users to change the over provisioning space in the firmware via software. Additionally over-provisioning serves as a cache partition for read/write operations and it's useful when a drive doesn't have dedicated physical cache memory, a full, so called "cacheless" Flash drive will slow down significantly. It is a legitimately good feature because the last thing you want is your data being gone sooner than before the end of a product's lifetime. But it should be clearly communicated or the packaging should inform what the actually usable storage space is.

[@JamieBarton1984]

you'd hope SanDisk would give you a bit of storage over the advertised amount to make sure the end user can actually use the full amount. but of course not!

[@justovision]

@@JamieBarton1984 Nobody makes a 132gb flash chip.

[@looks-suspicious]

@@justovision Nobody makes a 128 GB flash chip either, because we're talking about 128'000'000'000 bytes.

[@RobertHancock1]

All drives (that aren't garbage) have always done this, they just didn't include the capacity that was unavailable as part of the advertised size. That's false advertising.

[@9852323]

Firmware shouldn’t take up 4GB for something as simple as a flash drive.

[@dimitrioskalfakis]

the reason is that the wear leveling method used internally by the controller is coupled with some small percentage of space for replacing faulty future sectors with (move data around) in order to extend the useful life of the drive and its reliability.

[@vwestlife]

They've always done that, but the flash memory chips used to have enough excess capacity to allow some room for it while still providing the full advertised capacity to the user. Now they don't bother doing that anymore, and hope you won't notice.

[@katrinabryce]

Sure, but why not just print 250GB on the box rather than 256GB?

[@Bushougoma]

@@vwestlife This could be a change from the NAND manufacturers only manufacturing chips in standard sizes 128GB, 256GB, etc with no extra capacity. More capacity increases the silicon wafer size which increases cost. I wouldn't trust the no name flash drive manufacturers that appear to offer full capacity either. They may have simply turned off wear leveling to give you more space but at the cost of data integrity. This wouldn't effect most users unless the drive was heavily used. But if you did use the drive heavily file corruption without warning is inevitable.

[@KRAFTWERK2K6]

STILL tho..... you would expect there would be at least some standards everyone should have to fulfill... to give you AT LEAST 98% of what is promised. I simply don't take that whole thing as an excuse anymore.

[@SanderEvers]

@@Bushougoma True, and it's also a bit of Silicon Lottery, some might have a few dead sectors which would give them a "close enough" raw storage. Also these drives have gotten a lot cheaper, which well.. reduces their quality.

[@Kyuunex]

The key is in the "optimizations that are used for the performance and endurance of the product", I think you glossed over it. This is a valid reason. Flash storage is imperfect, memory cells die time to time. To avoid write errors when this happens, some part of the actual storage is reserved to compensate for the cells that died. Another reason for reserving some storage is wear leveling. Of course, the no-name brands don't care about the long term usability of their products, in fact I have a no-name flash drive that errors out when a certain cells of it is written to, probably because some of the cells died. Flash memory chips come in capacities of binary numbers, so, 4, 8,16, 32, etc. as an example have never heard of a 18 GB chip existing and used in a 16 GB drive with 2 GB reserved. Manufacturers refuse to advertise lower capacity with flash drivers but they do for SSDs, as an example, a 120 GB SSD more likely then not has 128 GB worth of memory chips, with 8 GB reserved. I hope this helps!

[@vwestlife]

That doesn't explain why of the two "128 GB" SanDisk drives I showed, one delivers its full advertised capacity while the other one does not. I can't imagine the one which actually gives the user 128 GB of available storage has no wear levelling or error correction while the other one which has only 124 GB of user available storage does.

[@oscodains]

Would it not be better to add extra gb if you’re going to remove it from user? So the result is still the advertised storage amount?

[@eDoc2020]

@@vwestlife The one with 128GB (128,000,000,000 bytes) of presented capacity does have hidden wear leveling sectors. The amount? 128*(2^30-10^9) bytes.

[@joshm264]

I've said it before and I'll say it again, the confusion between decimal and binary gigabytes is one of the biggest slight annoyances with computers, because in the end it doesn't really matter

[@johnps1670]

Could be the difference between a working or failing backup.

[@TravisTev]

I argue that with the size of today's digital media it scales up to a significant difference. If you buy a 5TB HDD (which are widely available for affordable prices) thinking it's in binary TB, you'll actually get over 460 GB less than expected. 460 GB is a big difference!

[@briannem.6787]

I think sometimes binary terabytes can sometimes be called Tebibytes (TiB) but I rarely see this difference actually noted consistently.

[@z1ej200]

linkin park

[@spongeyperson]

You're clicking on the partition, not on the actual drive in Disk Utility. This means, any one of these drives could actually have had a little bit more storage, but because of the way it was formatted, some of it is either unformatted, or being used for wear levelling.

[@schmudej85]

One thing that also may have made these companies bolder, and more willing to cheat us, is that people have grown so used to the way most operating systems measure these (via multiples of 1024) and expect to see a discrepancy. MacOS is, as far as I am aware, the only os on computers which measures decimal gb by default. Apple changed MacOS some years ago to use decimal instead of binary storage measurements. With Windows still measuring in binary, these companies likely figured people were so used to seeing differences between marketed capacity vs available capacity that a little more wouldn't even be noticed and, sadly, it seems they were largely right in the case of most consumers.

[@yellowcrescent]

I usually attributed the difference to GB vs GiB, since I normally don't look at the size expressed in base-10. Although I don't really use USB thumb drives very often. Out of curiosity, I just looked at some of my various storage devices. It seems that USB thumbdrives are usually lower than advertised. Fixed disks are always at or above. And SD cards veeery close. (1GB = 10^9 bytes, 1TB = 10^12) USB: Sandisk Cruzer 16GB = ~15.59GB, Microcenter 8GB = ~7.93GB, Verbatim 2GB = ~1.94GB, PNY 16GB = ~15.87GB SD Cards: Sandisk Extreme 512GB = ~511.9GB, Sandisk Extreme 64GB = ~63.9GB Fixed disks: Samsung 960 EVO 500GB = ~500.1GB, Samsung 860 EVO 500GB = ~500.1GB, Samsung 870 EVO 1TB = 1.000TB, HGST Ultrastar 18TB = 18.000TB, HGST Ultrastar 6TB = 6.001TB

[@stereophonicstuff]

We’ve all gotten so accustomed to this that the industry itself continues to perpetrate this “scam.” Personally, I’ve gotten so used to expecting that the new memory card or flash drive I’m buying is going to cheat me out of some storage, so nothing surprises me these days. I have a mountain of old and new SD cards that I’m quite tempted to test and see if the “actual user storage” is less than advertised, just like with these flash drives. I may just have to make a follow-up video to yours putting SD cards to the test!

[@SparkY0]

Imagine if a gas station redefined a gallon from 128 ounces down to "100 ounces", and then only sold you 95 ounces when charging for the full gallon. I don't think ant part of that would be legal.

[@johnathin0061892]

Actually... that is a thing, sort of. kzbin.info/www/bejne/mHSrnYWqf994b6c

[@SparkY0]

Storage manufactures should be required to print the useable capacity of their drive (in actual Bytes) somewhere on the package. It's not a big deal if it's in the fine print because yes, that's going to be a long ugly number, but printing the real capacity would take less space on the back than the multiple disclaimers explaining that their marketed capacity is a lie. Every other industry has their weights and measures regulated. A food company can't write "One Dozen Eggs - actual egg count less". They've been don't this a long time, but there's got to be some lawmaker or judge somewhere who is capable of understanding why this is all BS.

[@liucyrus22]

Should refer to how the baking standards are enforced and how eventually they just over provisioned to ensure that they meet standards.

[@HappyBeezerStudios]

Usable bytes in which partition scheme?

[@jondough76]

This has been the case for almost every storage medium forever. The raw size and the formatted size are always going to differ depending on the partitioning scheme and the format used.

[@proCaylak]

7:06 I had a "Norton Internet Security Netbook Edition" installer on the thumb drive that otherwise looks exactly the same. It is 1GB though(1,048,571,904 bytes). I have formatted the drive many times, but it still works like a charm for small bootable disk/storage purposes.

[@jigpu2630]

I'd suspect that this practice is more common on "large" "fast" drives. The cheapest way to get high capacity is to use NAND flash that stores multiple bits per cell (i.e. MLC, TLC, or QLC NAND). Unfortunately, this is slower than traditional SLC NAND that only stores one bit per cell. Sometimes, however, these higher capacity NAND chips allows you to operate a portion in SLC mode so that you can get lots of storage and fast speed (at least until the reseved space fills up) for cheap --- but you *do* give up a small amount of space in exchange.

[@fallwitch]

I recently bought a thumb drive off Amazon that was 5 Star rated AND "Amazon recommended". Not only was the storage one tenth of what they advertised the speed was one-thousandth of what they advertised. SMH.

[@vwestlife]

That's one thing that disgusts me. Amazon and eBay claim they have a team of experts working to make sure that counterfeit fashion items (clothing, purses, jewellery, etc.) don't get listed for sale, but when it comes to counterfeit electronics, they turn a blind eye to it, and Amazon even stocks it in their own warehouses and promotes it to the top of search results.

[@poofygoof]

usable storage will also be dependent on the filesystem used to format the disk. I guess us old timers are used to this stuff from the base2 vs base10 stuff with hard drives, but at least there they used to indicate mebi vs mega in fine print on the spec sheet.

[@ct1660]

This is common on SSDs as well, the SanDisk reporting as 124gb is due to storage set aside to account for wear leveling due to bad sectors over time. The larger the storage drive, the more it needs set aside. I’ve had good luck with Micro-Center USB drives, you do get the rated storage capacity when you check the drive on Disk Utility. However, memory chips are manufactured in powers of 2 and typically like being in pairs for optimal performance. Hence why there’s 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048 and so on.. It would require creating a whole new circuit for mismatched modules, say you have a 256GB drive and you pair it with a 8GB module for over-provisioning. It would cost a bit more plus a bit of latency when it does eventually have to replace bad sectors from the main chip.

[@vwestlife]

If the chips are manufactured in powers of two, then a true 128 binary GB drive would have a capacity of 137 decimal GB, so there's already 9 GB of overhead for wear levelling even if the full advertised 128 decimal GB is available to the user.

[@RussKnize]

This is pretty normal with flash memory these days. The base-2 vs base-10 difference used to be enough overprovisioning for wear leveling and write performance, but the firmware has knobs to increase the holdback to get better write performance and endurance. You are still getting all the memory, but you cant use it all for storage. Some of it is borrowed against the future. Like you said, most users wont care.

[@ram89572]

What we should be doing is pushing for transparency in their advertising. It's fine if they need to do that but they shouldn't advertise a drive capacity that they are only giving you in the most technical sense. I know when I buy one it won't be the rated capacity to me. Your average consumer does not know this. If they had to advertise it clearly with what is available to the user it would make everyone's life a little easier. They could then also advertise that they are using the remainder of the space for future longevity of the device to help you, the consumer, out. They advertise it the way they do because it helps them to advertise the biggest number they can

[@SianaGearz]

@@ram89572 Let's say we introduce a "guaranteed accessible capacity" rating that they all have to write on the package. This invites the sort of competition you don't want: manufacturers trying to cut down the size of hidden management area to give you a tiny sub percentage of extra bytes but reducing reliability and performance in the process. A competition on who can make the shittiest drive just to have a better number. Do you really want that?

[@ram89572]

@@SianaGearz You literally already have that. You already have a competition on advertising the biggest number for the least amount of product they can give. How does making them be transparent with their product in any way worsen your prospect at actually getting what is advertised

[@SianaGearz]

​@@ram89572 What you're missing here is that let's say they write "128GB", then all of these drives have exactly 128GiB or Terabit size flash IC in there, regardless of whether the usable area size reaches 128GB or not. There are no intermediate sizes, next smaller is half-terabit and next larger is two terabit. If they reduced the size of user area and increased the size of management area, it doesn't mean they are stealing bits and selling them on the side to someone else, they put all the bits in your product but repurposed them to improve performance or endurance of the drive. Normally it used to be the case that 7% of flash die space was enough for management area, this is the capacity difference between GB and GiB; but back then it was the era of USB2 drives out of which you don't expect much performance anyway. Also flash ICs used to be a lot simpler than they are now, so i'm not surprised the size of management area is growing. You also don't have an incentive today for the manufacturer to release a drive claiming to be 132GB in size, people would think it's weird and fishy. And indeed no chance in hell that would be a well performing drive on a terabit flash IC, though there's obviously a way to make it fit! "How does it worsen your prospect at actually getting what is advertised" - oh you would get exactly what is advertised in capacity, you just would get a shit performance since there are no formal performance and endurance standards or requirements at all. Just try to buy some off-brand drive like XLayer - you get full capacity advertised, you do, but at hilarious 2MB/s write speed over USB3, and it dies in a few weeks. Be careful what you wish for, that's all.

[@nickwallette6201]

@@ram89572 Because then ill-informed consumers start comparing drives based on the number of "real" accessible bytes, which pressures the manufacturer to trade more usable space for less long-term reliability. In the end, the vast VAST majority of consumers aren't nit-picking the last 2-3% of space available to them. They're just using it as a black hole of space until it stops working. Ideally, it'll stop working because it's full. People tend to understand that. Less ideally is when it stops working because the flash has worn out, because people tend to hold that against the manufacturer. So while "more transparency" would be a noble goal, people aren't educated enough on storage esoteria, and so it's probably better they just aren't asked to think about it, because they wouldn't have noticed anyway.

[@uxwbill]

I'm not sure I'd care to die on this particular hill, though to each their own. 🙂 This does have me curious (as at least one other commenter has suggested) if rather than over-provisioning the flash memory chip, they shave a little capacity off the top to cover those regions that will fail. Even at the kind of scale at which these things are made, I have a hard time imagining that it saved any amount of money to forego the spare storage area. I bought a bevy of those orange USB flash drives because I was tired of people at work "forgetting" to return my flash drives. While they are legit, a few of them had bad spots out of the box that their controller couldn't seem to map out with good spares. A few more have died with continued use. Looking inside suggested they were using a decent controller, with B grade flash storage memory at best.

[@JamieBarton1984]

more videos pretty please 🙏

[@accordinglyryan]

A wild uxwbill appears! I'm not surprised to see you here but hi lol

[@gazzmanp]

All of the ones that I noticed that were 'lying' were NTFS or exfat formatted. Others were DOS. NTFS has overhead. You are looking at the formatted volume, not raw capacity. They are probably all correct. It depends what file system you use.

[@HappyBeezerStudios]

Exactly, different file systems will have different capacities.

[@martytoo]

The tiny SanDisk is perfect for adding MP3 capacity in a car. My 2013 Honda had 2 USB slots and I used the one that could be read by the music system to store music for playback while driving. Unfortunately, our newer Honda and Acura won't both allow for memory disk storage and also allow for Apple Car Play or Android Auto. If you use the cell phone to car connection, you can't use the USB stick for playing MP3's at the same time. Those physically small drives are also a great way to add storage to a Chromebook.

[@sf-dn8rh]

Very few honda s got that feature. My 2008 accord with the upgraded stereo without GPS does not have USB access, only aux

[@mayzelel]

Thank you for keeping us informed and keeping them - manufacturers - responsible, these “little” things do matter

[@stepheneickhoff4953]

I'm an IT pro. I'd be interested in hearing about the exact "optimizations" that require them stealing 4 GB.

[@beerrox711]

Their wallets?

[@Vageta1999]

It’s called over provisioning. Without it the drive would wear much faster and fail much sooner.

[@EeekItsSnek]

All the "device not ejected properly" warnings at the end made me chuckle. 🙃

[@aktronics]

Brother loves cassettes so much went out and bought a flash drive that look like a cassette 😂

[@Markimark151]

SanDisk is so guilty of misleading their actual USB storage capacity for 64GB, 128GB, even 256GB! I bought a 64GB and 128GB flash drive on sale last year and they’re off by 5GB or 8GB less respectively! This is why I don’t trust them and even Richard from RTU hates SanDisk USB flash drives with a passion!

[@adamjohnson4311]

This is exactly the kind of investigative journalism we need more of. Looking forward to seeing that episode Turbo explained, on a PC you'll never see.

[@Kennephone]

I find it insane that high capacity flash drives can be made to be barely bigger than the usb plug, at affordable prices too. They make them up to "512gb" , and they cost less than $40. Remember the old days when a 256gb ssd had chips covering the whole board, and cost hundreds of dollars, but hey, at least you actually got the whole 256gb.

[@qixxxz]

The controller may turn off parts of the memory that is bad or use it to wear level and cash short bursts of data transfer. This is low level % offset compared to the scam drives that are 100 Terabyte drives for 10 bucks.

[@meetoo594]

Correct me if im wrong, but isnt the total available space reported by macos and others defined by the file system used? ntfs and fat will be different as will the unformatted ones which seem to report the correct capacity. Depending on the age of the drives it might be due to the wear levelling shutting out bad sectors as well.

[@vwestlife]

NTFS actually has less overhead than FAT32, and it only affects the efficiency of data storage on the drive, not the total capacity available to the user.

[@meetoo594]

@@vwestlife Formatting does take up a bit of space though and I imagine along with bad sectors due to wear leveling the inbuilt os disk tool hides the unusable space and only reports whats usable. I could be wrong though, never used Macos.

[@plan7a]

It, perhaps, should be noted that most of those tested seem to have less for a few different reasons. Firstly, they are formatted to a different file system type. (Almost all those tested which were 'not quite the amount stated' were either NTFS or exFat). Secondly, some of those drives have built-in 'security software' (or 'optimizing software') to help protect the files that are stored on them. (The new one you tested does say this on the packaging, so I'd assume some of the space 'missing' is used for this). This does *not* necessarily, however, totally prove anything - as *some* of those you tested are rulebreakers for this. Thirdly, it is also something to note that some CD/DVD (etc) discs are pretty much the same when it comes to actual usable space available. Some big branded discs can vary also (based on my experience) from those which are more 'no name' branded ones. [I've had issues where burning a duplicate copy of a disc can sometimes fail because the amount of space available differs, albeit just a tad in most cases (from disc to disc); even though they should be the same as each other, surely]. It's also the case when software *can* allow slight over-burning of discs too. Some will, some won't. Of course hard disks/drives can also vary to some degree (with the 'security' or 'optimizing software' also). Perhaps part of this comes down to making the devices usable with older legacy devices as well as newer ones? But this is just me guessing. Just some thoughts on this one; sorry the comment is a bit long.

[@hgbugalou]

I really wish tech youtube would unite with videos here to force Amazon to do something about those fake drives. It infuriates me so many are listed and Amazon should try a lot harder.

[@daveys]

Just don’t buy anything like that from Amazon Marketplace, but I agree that Amazon is no better than Ali in the advertising of dodgy USB media.

[@rzeka]

2:25 classic PR tactic. "If they point out a flaw in the product, just tell them it's an ~optimization~!"

[@dlarge6502]

The missing space is likely used for wear leveling. Flash controllers can be programmed to partition the flash chip in a variety of ways. This partition scheme is above the level we have access to so we cant modify it. We have access to the user partition, which is where our data goes, including the filesystem and our partitioning. Some older drives came with utilities that allowed users to edit the higher level of partitioning which gave amazing god powers over how the flash drive behaves. It was possible to create a read only user partition, or two user partitions one read only and one writable. But it was also possible to reduce the size of the user partition and create an ISO partition alongside it. A CDROM or DVDROM iso image uploaded to this iso partition would then appear as an emulated read only usb cdrom drive with that iso! The drives that were released with such functionality were called "S3" drives and they can still be found as new old stock, you might want to hunt one down and make a video about it. Many drives from different manufacturers supported S3 so its possible some of your older drives may already do so, with the right tool you can then play with the flash chip partitioning. My old PNY 8GB does support S3. Unfortunately this functionality is now lost to us because we apparently dont need to use it according to marketing departments. This is VERY annoying as the ability to create a read only USB drive is a feature many geeks wish they had and the ability to create a usb drive that also provides an emulated CDROM with a user supplied ISO file... WHO THE HELL DOESN'T WANT TO DO THAT?? The functions are still there, they are however used to set up the drive in the factory. You can order USB flash drives in bulk that provides read only application installers or OS installers, well this is how its done. Besides the S3 drives the only way us little people can mess with these god powers is to identify the controller on the thing and download tools from nasty looking russian websites and forums. SSD's use some of their capacity for wear leveling, but they are advertised with the available capacity! Here someone thinks they can get away without updating the packaging.

[@cybernetennis]

you are comparing NTFS volumes to FAT? Its going to come up different due to allocation size too,....

[@vwestlife]

NTFS actually has _less_ overhead than FAT, but either way that's irrelevant because the overhead only affects the efficiency of the user data stored on the drive, not its total capacity.

[@cybernetennis]

@@vwestlife yeah so I guess the discrepancy comes for how much space is allocated for wear leveling?

[@jimellis5604]

It's understandable to call a 124gb "user available" drive 128gb. If they called it 124gb it would cause lots of issues with sales due to the fact that we all search for standardized capacities. 4, 8, 16, 32... Someone has to use these flash chips that don't have extra space for error correction, wear leveling, and such. If they made it all available to the user, it wouldn't be a reliable device. All they really need to do is make sure it is VERY clear and upfront that the actual available capacity is what whatever it actually is. For example lable it 8gb, but right next to it, VERY clearly, write "7.6gb usable" A lot of those generic drives with the full capacity available might corrupt your data because they aren't reserving space for wear leveling, replacing bad blocks, and all that good stuff.

[@peterripson]

I noticed these discrepancies a long time ago and just figured it was the price of admission, sadly. Some of my older ones that cost me a fortune back in the day (and even a couple not so old) are starting to fail us one by one, so there is that to look forward to as well.

[@jordanch68]

A lot of these just repackage bulk drive modules from China, the only difference is cosmetic. Lately I've taken to getting them from the various China sites, testing them, and repurchase from vendors that check out. I would take issue with the explanation from Sandisk because the unformatted capacity is always more than the formatted capacity which is used by the user. The reduced capacity available to the user has nothing to do with the actual capacity of the memory chips. There should be an unformatted capacity and a formatted capacity, if the unformatted capacity is less than the advertised capacity, it's false advertising. That's why I don't buy their explanation, they say "capacity available to user less" but the unformatted capacity is not available to the user so their explanation doesn't apply. I wonder if manufacturers might get around this by simply setting the max capacity value in the module firmware, just to report the value not really doing anything useful.

[@AnonymousFreakYT]

When storage device manufacturers were making spinning-rust platters whose capacities were essentially arbitrary based on non-binary things (physical size of platters, number of platters, density of data,) this "binary vs. decimal" made sense. But for flash storage, where it is nearly always sold in "powers of two" capacities? It's ridiculous to even *attempt* to use decimal. When one flash drive is 2^40 bits, and another is 2^42 bits, those should be 128 GB and 512 GB respectively. I understand there is reserved space for wear leveling, but just tell us how much on the packaging; or over-provision, making it 1.25 * 2^40 bits or something. (six chips instead of four, or whatever over-provisioning they want.)

[@tangyorangegames7488]

Hehe quite informative as usual. Love this channel. I think the cassette tape usb is quite cute too.

[@gorillazfan85]

I hate how pretty much no drive I buy has its actual listed capacity when they very easily could correct the proper way.

[@MrAlan1828]

Just a fun fact. A 3.5 floppy is actually 2mb in capcity but after formatting you get 1.45mb of usable storage but you can still format to an uncommon 1.6mb but its not reliable and not many computers at the time can read write to it

[@SilverX95]

It really depends on the type of storage that they used. like in my case with a MicroSD card from Silicon Power that uses 3D NAND, some of the storage would be used for maintenance, like wear leveling and bad sectors. you also need a factor in the file system too, like NTFS and FAT and also the partition table like Master boot and GUID partition table and others there's a lot of them. though i do wish they would just add the extra space for that's stuff so you get the full size.

[@SianaGearz]

Where is the extra space gonna come from exactly? A 128GiB die and a tiny little 8GiB companion die just to make that space? That's dumb, that's extra cost that nobody needs, not you not the manufacturer, those smaller capacity flash sizes can no longer be made economically. All flash die sizes are power of two in bits, so a 128GB disk has a Terabit flash in there. If they can fit the scratch space, wear levelling space, firmware etc in 7% of that space, you get a true "128GB" drive. If they choose to expand the size of the management area to improve performance or reliability beyond that, you get less space. It's not like the drive manufacturers steal bits from your drive to sell them on to someone else later.

[@nickwallette6201]

@@SianaGearz Exactly. You got what you paid for, some of it just goes to "business expenses" of running the drive. These aren't small drives anymore. It hurts when 128GB of data just vanishes due to a hardware fault, so delaying that and trying to prevent it are worth the small cost in accessible storage.

[@horusfalcon]

A lot of this "cheating" is because some space is used by the operating system for formatting and cache-like operations. As drives grow in capacity, the need for some "overhead" is a real thing. Some manufacturers do a better job of communicating this than others do, apparently.

[@Ale.K7]

After years and years of the dreaded "up to" in transfer speeds, now we have "up to" in storage capacity.... I wonder what will come next.

[@chounoki]

Generally all cheap USB drives are made from defective NAND flash chips. Those defective NAND flash chips are born with bad blocks in the factory, but since the bad blocks are usually just a very small percentage like 2% or 4%, it would be a waste to throw away the entire gigabytes chip, so they are made into cheap USB drives with the bad blocks masked out as unusable. That's why almost every single cheap USB drive comes with just a little less capacity that the printed spec on the package.

[@BobBell808]

I wonder the reason as to why there are these discrepancies. I would guess that there is a built-in failure rate when assembling these large arrays of memories. Because these failures are not exact, they would have some sort of range. The manufacturers would then need to decide what amount of failure would be acceptable; past a certain percentage would reject the drive. As a math geek, I ran a spreadsheet to get the percentages of each flash drive. The range was from 96.45% (PNY 64GB) to 101.25% (PNY 8GB). Interesting that the other two PNY 64GB you tested were 96.81% and 101.48%. In fact, all the other drives fit in between the PNY flash drives. Branch Education has a fascinating video on how NAND Flash memory works. It really blows my mind.

[@interlace84]

They can say they're doing some "overprovisioning" to maintain performance but since you're unable to verify that in any way-- pretty shady business practice.

[@robine5280]

Wouldn't "over"provisioning also indicate they would add some storage over the advertised (and usable) amount to deal with cell wear?

[@segarallychampionship702]

My Samsung USB drive also gets really hot, and I think that's just due to the metal case that better conducts heat combined with the increased heat from the flash chips due to running at higher speed to meet/make use of the USB 3.0 spec

[@KRAFTWERK2K6]

yeah Some get really hot and others you can leave plugged in for hours and they won't get warm. Too bad you never really know. These mini SanDisc sticks that kevin here was showing, really DO tend to get really hot.

[@dwarf365]

@@KRAFTWERK2K6 I have the 64gb version of the red and black SanDisk and it does get extremely hot when filling it up or when used for an OS drive for a few hours.

[@JeffCD77]

I get anxiety everytime i see a flash drive being yanked out without it properly being ejected because nowadays MacOS screams at you for doing that. **And apparently older versions do too.

[@joshm264]

Mac OS X/macOS to my knowledge has always given you a warning when you unplug it without ejecting, but in 10.9 they changed it from a giant red pop-up to a notification in the corner

[@vwestlife]

Keep watching...

[@londongaz2]

​@@vwestlife that bit at the end made me smile!

[@rdoursenaud]

As long as no data is being written, it's fine…

[@jhwilder70]

I thought I might have been the only person cringing each time he did that until I scrolled towards the end of the comments. I feel your pain @JeffCD77. That bit at the end had me laughing too.

[@Jdbye]

They're technically all correct. The flash chip inside is the correct capacity, and that's what you're paying for. It's just a matter of how much is set aside for wear leveling. Flash drives with NO space reserved for wear leveling can have a very poor life expectancy, so it's an important feature. It can also have an effect on performance. I would prefer if they advertised the usable size instead of the raw size, but less tech savvy people would expect to pay less for it compared to other products that advertise the full raw size or have no reserved space, and nobody would buy it. Making such a product a non-starter and not worth manufacturing. So companies were forced to make the choice between not introducing features that improve the lifetime and performance at the cost of sacrificing some space, or to "lie" about the size, and they chose the latter. This is why older/smaller drives tend to advertise the "correct" capacity more than newer ones, the technology was more simple and often had no wear leveling at all. Which is a bad thing.

[@vwestlife]

The flash memory chip have capacities in powers of two, thus a true 128 binary GB drive would have a capacity of 137 decimal GB, so there's already 9 GB of overhead for wear levelling even if the full advertised 128 decimal GB is available to the user.

[@jort93z]

It's quite normal that some is used for optimisation(wear levling) and such. Most SSD's will not be sold as 256 GB, but as 250 GB or 240 GB They do generally have 256 GB of storage on there, but not all of it is available as user storage. Some USB flashdrive manufacturers seem to just advertise the raw number.

[@gur3n6089]

Most consumer SSD nowadays do advertise themselves for capacity like 256GB, etc. Enterprise SSDs are more likely to use other numbers like 280/960, etc. But enterprise drives are a lot more expensive and the brands typically don't sell them at retail.

[@segarallychampionship702]

240GB and multipliers are generally DRAM-less SSDs (ie. without cache). SSDs with DRAM are either 250 or 256GB.

[@gur3n6089]

@@segarallychampionship702 TIL, didn't know this correlation before. I was looking at Optane drives while typing that comment, which don't really need DRAM caches.

[@dutchbeef8920]

4 gb for optimisation? Come on man

[@eDoc2020]

@@dutchbeef8920 In reality 4GB is very little overhead for 256GB of raw storage. Having extra space for the wear leveling algorithm to work with can greatly reduce the amount of write amplification and thus make the drive last _much_ longer. Enterprise SSDs (at least a few years ago) came in sizes like 200GB instead of the 240 of a consumer counterpart.

[@acerbt]

I think part of this has to du with 2 things, which is the act of formatting the drive as well as how windows and mac treat storage when it comes to displaying it. Mac systems display the storage correctly, that is to say the units of measure like gigabytes line up with what's displayed. Windows on the other hand, measures in binary storage values, so 1024 instead of 1000, which means that gigabytes should show as gibabytes, megabytes should display as mebabytes, etc since the calculations are done with bass 2 values instead of bass 10. The problem comes up with the displayed unit of measure being wrong on windows machines. Also, formatting always drops some amount of usable storage. What makes this even more confuzing is that storrage manufacturers don't define which scale, binary or bass 10, they use to measure storage sizes, but ram is always written with the correct size.

[@Akotski-ys9rr]

Whoever I use any storage device I pretty much expect it to have a few gigs less than advertised

[@LightTheUnicorn]

While I'm sure the physical flash is the advertised size and some is missing for wear-leveling, that really should not be the point, and I fully agree the advertised size should be the size you can actually use. I've gotten so jaded with modern storage devices being so penny-pinched though, that frankly I expect I'm getting screwed over in some way whenever I buy basically anything like this.

[@weasel2htm]

I would like to know what those optimizations are. I wondered if it was for some fault tolerance, if parts of the chip started to fail, data could be moved to a previously unused part of the chip, reading the comments have confirmed that for me. But really, on the 256GB drive, you're losing less than 1%, I will gladly trade that for some extra reliability. That being said, it would not kill them to advertise the usable capacity, but that would make things a bit more confusing for consumers. Besides, companies not stretching the truth on their products, I'm pretty sure that breaks some kind of "bro code" among them!

[@james-5560]

I hope this videos gets a million views and gets picked up elsewhere so the manufacturers are encouraged not to cheat us. Would the same apply with solid state hard drives?

[@KRAFTWERK2K6]

I swear, selling us USB sticks with LESS capacity than advertised is like the digital version of buying potato chips with 20% chips inside and 80% air.

[@b2gills]

​@@KRAFTWERK2K6 you are aware that potato chip bags have exactly the right ratio of air to chips so that they have the fewest broken chips. There was even a Food Theory video about that.

[@hfric]

The thing that all pointed out is , that depending on the drive manufacturer or the model ... some drives partition the chip, to have cache to fake speeds , done mostly on drives that can overheat... fine example are Samsung Bar Plus USB 3.1 sticks, that use theirs aluminum case as a heatsink ... giving you real 126\256\512GB space on them , 100MB\s writes 300MB\s reads (even 400MB\s) ... compared to theirs plastic counterparts that overheat and slow down ... but this Samsung Bar Plus can get really boiling hot ...

[@McRocket]

Thank you for this. And I think your theory (at the end) makes sense. ☮

[@matt.604]

I think ice cream manufacturers are doing the same damn thing.. same size package but filling it up less and putting the volume in very small print. Oh and also adding a lot more air into the ice cream.

[@AlguienMas]

6:15 Hey I have that exact same drive, but is 128 gb. It´s a shame it suffers from that anomaly

[@RPKGameVids]

I have the orange one too, but it has a little holographic sticker on it saying it's 32GB. Not sure what the actual storage space is, but I'm sure it's not that far from what the sticker says.

[@abluestreetscene]

7:25 Am I misreading it or that USB is using negative space?

[@50shadesofbeige88]

What you're describing here is space that's been designated for static wear leveling. This is especially common on older USB drives. You don't see it as much on the newer ones because the newer controllers can do dynamic wear leveling that takes into account the available free space. If anything, I'd be upset that you may be paying a premium price for a flash drive that uses an older controller.

[@igorszamaszow171]

To be honest, these discrepancies do not bother me that much. I always try to leave around 1/8 capacity empty anyway

[@prodbyfaith]

Where were these drives last formatted? macOS tends to add a 300 MB partition to the start of every GPT drive, so that could explain a lot, along with overprovisioning. You can't actually properly measure the capacity of a drive the way you're doing it because of all kinds of little differences between systems.

[@donnierussellii4659]

The real kicker here is that the loss of capacity is due to defects in the memory, which are scanned at manufacturing, so they know what they are selling. Binning is used extensively in consumer electronics, such as when you buy a cheaper CPU which is actually partially defective more expensive CPU.

[@mrnmrn1]

They use standard sized flash chips, but for the sake of reliablility, they are not using the whole capacity of it, so they have some spare space. If the flash chip starts to wear out, there are some spare space they can replace the bad (worn) sectors with. So this is actually a good thing, data corruption is less likely on such drives. A colleague had a 16GB SanDisk drive, that actually made itself read-only, due to the wear of the flash chip. All the data remained intact, but it was impossible to write anything to the drive, or deleting partitions/formatting. SSDs do the same, it's nice that they are implementing the same thing in high capacity thumb drives. It's better to have 4GB less space than loosing 128GBs of important data. Of course, noone should store 100s of GBs of important data on thumb drives without backup, but most people just don't care. And using thumb drives (or SSDs, SD cards, any flash memory) for long-term storage is the biggest no-no! The data retention time of the more modern, very high data density flash chips is less than 10 years, usually not more than 5 years is guaranteed with a brand new drive, but the data retention time declines with wear, and depends on the chip temperature during writing and storage.

[@Aeduo]

I'm one of the last to be on the side of a business but this does make sense. Each chip will be ideally some round binary number but they'll come with faults, maybe extra data to manage their functionality and also reserved space for faults which will develop over time. To have all that while having the precise binary gigabyte would require making odd size memory chips, which is certainly possible as these are purpose made all in one chips. It's probably just cheaper not to have the extra addressable pages and having a nice "square" flash memory.

[@bf0189]

If they needed redundancy then why didnt they just state the actual usable space and the reason why clearly in the packaging?

[@lmoore3rd]

Due to silicon wafer manufacturing, NAND Flash doesn't come off the line perfect every time. It comes off the line good enough. Costco was even selling some weird SanDisk 400 GB micro SD cards which we know are likely 512 GB chips that didn't meet the quality bar. They're the factory seconds offered at a discount for value pricing.

[@adammars1438]

My dad fell for this scam. He claimed a flash drive (similar to the ones shown in this video) would hold the advertised “2 TB” worth of storage. Running it through a test would find that it would hold a max of 8GB. Lessons were learnt that day.

[@rivards1]

It's funny to call even a half-gig shortage as "closeenough". That's 500 megabytes. My very first PC hard drive was only 220 meg. Now that's a rounding error.

[@rayphoenix7296]

I always accepted the fact that USB memory sticks never have the full capacity advertised. I own two Seagate hard drives and none of them have the exact amount of storage space advertised. I only like them because they can still hold a lot and they work without failing.

[@sophiespaan2995]

I am pretty sure the overhead on faster flash drives is the high-speed cache baked into the storage chip, as well as some form of buffer for dead cells, or even error correction. I'm pretty sure that the USB 3.x drives just have somewhere near 1/16 the capacity as a faster cache or something, and all the companies buy from the same chip manufacturers.

[@Kane26510]

Both of the USB 3.0 drives were less than advertised. I wonder if the required 3.0 protocols use some storage space, but trade off in faster functionality. You are correct, I know very little of computing - just throwing out ideas. Good video, as usual.

[@lightbringer502]

This video feels like the "Busting Solar Calculators" spiritual sequel. I love when you put technology companies to task!

[@paper_gem]

I have a USB stick that's plugged into my computer right now. It's advertised as 8 gigs but only has 7.25 gigs available to me. Yeah, we're getting hoodwinked.

[@VOLTRONDEFENDER4440]

VWestlife try formatting the drive as I found out with those drives theirs a few gb thats unallocated not formated partition

[@yurriaanvanduyn]

I love the build-in features of Sandisk and Samsung to move around storage when parts of it is faulty, but don't deduct it from the advertised capacity. I mean, then they should create a 132GB drive, reserve 4 for the product's longevity and sell it as 128GB...

[@nylint]

i remember when sandisk - trying to get, i think it was win8 logo certification - silently started manufacturing flash drives that couldn't be used for booting. they would then delete threads made on their forums by people unhappy with the undisclosed change. good times! when it comes to sandisk i don't know about "well-respected" at least in my house. pny...i've just had bad luck with as far as reliability. samsung flash drives have been pretty good to me over the years.

[@coondogtheman]

I've had tons of memory devices say formatted user storage less. My first MP3 player had 32MB that's MEGABYTES of space built-in internal memory but when formatted it was 31.06 MB which is enough for maybe 5 songs depending on the bit rate. the 64mb multimedia card I bought back in the day which formats to i think 62mb. sadly it no longer works because I just tried it. Worked for a little with jumbled text and then I tried formatting it which failed.

[@2000vitaly]

Also keep in mind, that the file system used to format the drive affects the capacity. Most of the "cheating" drives were formatted with NTFS which is not natively supported by macOS, as far as I know.

[@wilkes85]

I did notice this with a 128GB SanDisk. I just thought it was part of the "unusable space". It always bugged me that thumb drives don't actually have the usable space they're labelled as. It would be like selling C-90 cassettes that are actually 81 minutes. Even the cheapest bargain basement "low noise" tapes that are just re-slit factory reject video tape don't try to screw you out of some recording time.

[@TryItAgainTomorrow]

Interesting that the Patriot XT appeared to -2.1MB (note the negative) space used... :D

[@rubberduck4966]

This is because of the "Working Area" the USB Storage Controller need for their work. There are the advertised Capacity in GiB but the working-area is not direct useable.

[@kaitlyn__L]

I notice they're all binary numbers of deficit, or a combination of two slightly smaller ones. Even a couple you gave a pass seemed to have 256 or 64MB set aside, there's also a 768MB deficit which is 512+256. The 6GB one is a bit more egregious but that's 2+4GBs so the same deal. I'm totally fine with manufacturers setting-aside some of the memory cells as backups for if others fail prematurely, but they shouldn't advertise it as being higher. Eg I use quite a few 112GB SSDs, which is clearly a 128GB drive with 16GB reserved to ensure it's reliable enough. But I don't mind because they're advertised as 112, not 128.

[@eDoc2020]

My _oldest_ flash drive is unique in that it doesn't cheat at all with its capacity. It's advertised as 10 megabytes and it presents exactly 10240 kilobytes to the OS. This drive is so old it's a Sundisk, made before they changed their name to the more familiar Sandisk.

[@bestage9429]

I have two Kingston DataTraveler 100G3 32GB flash drives and I had noticed the actual capacity on both of them is something like 30.9GB (that's in decimal base, not in binary base). Additionally I've noticed if you write large files to them, they slow down to an absolutely crawling speed quite shortly, as the thing gets hot inside. The way it feels, it seems like the plastic shell isn't directly bonded to the flash controller inside to dissipate the heat away quickly enough. It does live up to its advertised read speed rating of 100 MB/s however. These are the first flash drives I've owned where the actual capacity was less than the formatted. Previously I owned a SanDisk Cruzer Blade, Cruzer Switch, a "Memorex" which is old enough to still have an indicator light, and a generic no-name real capacity 8GB flash drive. That generic 8GB drive has actually failed after quite a number of write and erase cycles, it's the first time I've had a USB drive fail on me. It still writes data fine, but if I store a file on it, leave it away in storage for a couple of months, then come back and plug it in again, I will find that the files stored on the USB drive have become corrupt / "bit rotten". I mainly used this USB to do OS installs, and eventually after a few years of use I would have to re-write the OS install image on there after a few months of sitting in storage, because the installer would say the files were corrupt, during the setting up.

[@smoothchills3072]

My recently bought legit SanDisk Ultra 128GB A1 microSD card has 127.9 decimal GB. Just 100 MB less from 128 GB. It's exFAT formatted. It isn't a USB thumb drive but compared to 4GB less, this is almost the advertised capacity.

[@dickbrocke]

Many manufacturers have a system which scrapes little chunks (1% to 4%) of raw memory from the chip before it is mounted on the component board. With all of these small portions of raw memory saved up they can then make more flash drive chips for the same price. It's a bit like cheese producers who scrape small portions (half a percent or so) of raw cheese off one un-moulded block ingredient so that these scraps can be added to another block. They save millions of dollars this way.

[@SianaGearz]

Thing is Flash storage media that isn't exactly power of two byte size does NOT exist. Some of the capacity (varying size) is used for wear level management. There are numerous SSDs where you get a tuning utility access where you can set your own wear levelling area size, usually it's just north of 7% of flash capacity, corresponding to the GiB and GB difference, hiding in the shadow of the unit conversion. If they changed the size of wear levelling area, they haven't done this in order to cheat you out of a little bit of product and sell those extra bytes on the side later; odds are they figured out the existing smaller size of wear levelling area was not quite enough to achieve desired performance and long-term reliability. So what is a manufacturer to do? Write actual capacity, guaranteed accessible capacity? Then they lose out against 30 other manufacturers who don't do that. I would gladly pay extra for a drive which has a few gigabytes more dedicated to checksums, wear levelling and reserve pool, because if i'm gettting 128gb, it's because it's not too expensive and the next smaller size which is half the size might not fit my use-case; not because i need every single byte on there. Most people like comparing meaningless numbers and thinking they got more for their money, which they really didn't, the real underlying flash size is the same one way or another. It would also invite competition among drive manufacturers, who can make the slowest, least reliable drive just to deliver a small percentage extra capacity, and you likely don't want that.

[@neelsmostert]

What is boils down to, is how much money the manufacturer wants to make, and what they feel like on the day the drive is designed. Sandisk being part of Western Digital would for instance use once unit of measure and Western Digital would use something totally different for their hard drives.