Seagate Offers Refunds on 6.2 Million Hard Drives

An anonymous reader writes "Seagate has agreed to settle a lawsuit that alleges that the company mislead customers by selling them hard disk drives with less capacity than the company advertised. The suit states that Seagate's use of the decimal definition of the storage capacity term "gigabyte" was misleading and inaccurate: whereby 1GB = 1 billion bytes. In actuality, 1GB = 1,073,741,824 bytes — a difference of approximately 7% from Seagate's figures. Seagate is saying it will offer a cash refund or free backup and recovery software."

Direct Link to claims

[micksam7]

File online [harddrive-settlement.com] [no cash, just software]

Mail-in [harddrive-settlement.com] [cash or software, cash claim only if bought before 2006 & you have proof-of-purchase. 5% of what you paid]

SI units

[Anonymous Coward]

1 GB (gigabyte) = 10^9 B
1 GiB (gibibyte) = 2^30 B

RTFM

[Revotron]

Do your research - your point is pretty much ass-backwards. The manufacturers are quoting their sizes in gigabytes, which are SI units defined as 10^9 bytes. A gibibyte is the familiar 2^30, 1024MB unit that we all associate as being a gigabyte.

Definitions

[mduke]

IANAL, but I think the reason they lost is not based on whether 1GB is decimal or binary but because they did not specify the system they used to count it. If they said it was 1GB in decimal so 1GB = 1000MB and made that clear, then they probably would have been ok. But since they did not, 1GB = 1024MB was easier to demonstrate as a better, more common, and more readily accepted definition due to the way it was shown in the OS, and there was nothing on the packaging to negate this. So make sure if you use numbers, you say exactly what they are supposed to be.

Re:Direct Link to claims

[houstonbofh]

cash or software, cash claim only if bought before 2006 & you have proof-of-purchase. 5% of what you paid

The mail in form also allows you to use your drive serial number as proof if you do not have proper documentation.

Re:wow.... are you clueless!

[mrbcs]

Can't read can you asshole. I said I stopped buying Maxtor over 4 years ago.. that would be 2003. If you read this: http://www.seagate.com/ww/v/index.jsp?locale=en-US&name=Seagate_Technology_To__Acquire_Maxtor_Corporation&vgnextoid=1e8a814fef83e010VgnVCM100000dd04090aRCRD [seagate.com] Segate bought Maxtor 2 years after that.

Re:RTFM

[Ledsock]

Do your research - your point is pretty much ass-backwards. The manufacturers are quoting their sizes in gigabytes, which are SI units defined as 10^9 bytes. A gibibyte is the familiar 2^30, 1024MB unit that we all associate as being a gigabyte.

Actually, 1 GiB=1024 MiB. That's the whole issue of this case. MB!=MiB, as with kB and KiB, and GB and GiB. The difference between a GB and a GiB is roughly 6.87%, yet when you hit the TB/TiB level, the difference is roughly 9.05%. The greater the prefix, the more the inconsistency between the two units of measurement. I view this case as preventative action for the soon coming terabyte and tebibyte hard drives. As sizes grow, so do our losses (although, technically, they are advertising correctly, and the OS makers are using improper notation).

Re:Ahh, another valueless settlement.

[__aajfby9338]

Remember the "welp we had a glue factory fire so prices skyrocketed!" bullshit? Special glue just for memory ICs - and that scaled exactly with capacity? Yeah, that "glue factory fire."

That was a fire at a factory which made the epoxy resin used to encapsulate ICs. This wasn't "special glue just for memory ICs"; it was the black plastic stuff molded around each IC on the SIMM (or any other kinds of ICs with plastic packages, for that matter). Without that plastic overmold to protect the bond wires and support the leadframe, the ICs can't be handled, shipped, soldered down, etc. That fire messed up the whole electronics industry for a while. I'm not saying that the memory suppliers didn't gouge anybody (I have no information either way), but the resin factory fire really was a big deal. It caused problems at my company at the time, which made ICs used in hard disk drives.

Re:Much like a RAM settlement offer in the mail

[Hunter-Killer]

The catch for one of the most recent DRAM settlements (pricefixing; Rust Consulting) was that you had to purchase memory directly from the manufacturer. If you were a consumer, this was unlikely unless you bought directly from Micron/Crucial. I put in a CS ticket with Crucial, and received a copy of my invoices for the desired time period (about $400 worth). The settlement terms was compensation on a pro-rata basis; given the amount of memory sold during that time period vs the settlement sum, I believe it worked out to about 10% for Crucial. Still waiting on my check, so I can't confirm.

Re:It's not a longstanding history

[Anti_Climax]

Your description seems to be a hybrid of the tragedy of the commons and the Red Queen [wikipedia.org]

I say we call this hybrid theory the Tragedy of the Queen.

Re:Seems Silly to me

[absoluteflatness]

Drive manufacturers are in a position to make the much easier fix though. Changing OSes to report base-10 sizes, or to keep the existing sizes with the *iB notation requires changes from every OS manufacturer, and suddenly leaves them inconsistent with older versions of their products.

Whereas, for storage companies, it's a simple matter of changing the labeling and packaging. A switch to only using the base-2 sizes (my personal favorite) would also probably mean that the drive companies would start subtly altering drive sizes so they wouldn't be selling "85.7 GB" drives and would instead align neatly on a round GB number (which is actually incrementally harder to do now, because drives already are organized with power-of-2 blocks). Otherwise, they could simply list both sizes on the packaging, or include something to the effect "your computer will report this drive as having a capacity of X".

Really, the problem now with the MiB, KiB units is that people in general aren't aware that these are any different than MB and KB, and would likely only increase confusion. Another unit distinction that's still causing confusion is illustrated by your last line (comparing GB's with Gb's).

GB is gigabtye, where Gb is gigabit. So it's very easy to compare these, take Gb's and divide by eight: ta-da, GB. Another confusion taken advantage, this time, mostly by the Internet industry. Modem companies quickly jumped on the terms "56K modem" and such, which helpfully obscure that the K is for kilobits (base-10 again), of course until your computer reports you transferring data at a maximum of something like 5 KB (not to mention physical line limits that further decreased the actual maximum).

Really, computer-related industries seem to like to sow confusion in the market. The distinctions don't matter as much with increased capacities (even though the distinctions themselves increase in size). Take as an example, say you have a file your computer reports is exactly 1.41 MB in size. Ideally, this should fit on a 1.44 MB disk (filesystem usage of the disk aside), but that 1.44 MB is really 1440 KB, where KB is the base-2 unit, or only 1.40625 "standard" MB.

Re:What a crock

[julesh]

What a crock. Anyone that knows enough about computers to know that GB, MB, and KB are usually base-2 should also know enough to check whether the HDD measurement is in base-2 or base-10. Non-computer people would probably assume that they are base-10... or, more likely, merely that the bigger the number, the better. In my experience non-computer people have difficulty distinguishing between hard-drive space and RAM. Saying that they are somehow miraculously able to distinguish between base-2 and base-10 measurements is ridiculous.

To the average person, the distinction between base-2 and base-10 is meaningless, yes. That doesn't mean, however, that they aren't being cheated. Their interface with the computer, when they examine a file, will tell them how large it is using base-2 units. Disk space requirements on the back of software packages are written in base-2 units. Everything they see is in base-2 units, so this is how they estimate their requirements of disk size. And then they find out that the disk is being sold using different units.

It's a confusing situation, and the disk manufacturers deliberately switched in order to take advantage of it.

The Kilo-, Mega- and Giga- prefixes are always base-10 in SI. The IT industry should come up with different terms. Misusing them was a mistake in the '60s and it is a mistake now.

While I agree with you, and try to use Ki etc myself, I don't think this is a problem that can just disappear like that. People are used to thinking in terms of 1GB as ~1.1*10^9 bytes. They might not realise that they do, but they do. Changing perceptions is a long and slow process, and software manufacturers (the only people who can realistically change these perceptions) are reluctant to start because they fear confusing their users. They're probably right.

Re:Think this will set precedent?

[darthflo]

Since your post is written with about as much intelligence as one'd expect from a tree stump, I doubt you are going to grasp anything at all, but to try and help you anyway: Read what the U.S. gov't has to say about it [nist.gov]. If that's too dry for you, this wikipedia article [wikipedia.org] might be interesting, too.

Re:Seems Silly to me

[RzUpAnmsCwrds]

Note: The following post uses SI prefixes (e.g. GB=10^9 bytes, etc.) correctly. Binary prefixes are used when appropriate.

Every operating system, whether it be Windows NT, XP, or Vista, Linux, FreeBSD, or Solaris, states that 1Kb = 1024bytes, 1Mb = 1024Kb, and so on.

WRONG. Use a modern Linux distro. You will find that many tools either use the binary prefixes or use SI-standard prefix usage.

Why rewrite all software, and god forbid, patch all old software going back however many DECADES into the past to implement this change, when harddrive manufacturers could simply start labelling their drives correctly?

Because HDD manufacturers ARE labeling their drives correctly. "Giga" means 10^9 in SI. It always has and always will. The computer industry usage has never been correct.

The disparity only grows as we go up in prefixes. 1TiB = 1.099TB. 1PiB = 1.125PB. 1PiB = 1.153PB.

Moreover, the non-SI use is ambiguous. A "1.44MB" floppy is neither 1.44MiB nor 1.44MB, it's 1 440 KiB. A "650MB" CD is 650MiB, but a 4.7GB DVD is 4.7GB.

GiB is NEVER ambiguous. If you want to keep using the power-of-two units, use the proper prefixes. IEEE, NIST, and the IEC encourage it.

I don't see why hdd manufactureres are the ONE single exception to this long standing rule, and SI units be damned.

Hard drives, flash storage, DVDs, Blu-Ray/HD-DVD, transfer rates (8Mbps = 8 000 000 bps), and everything else uses the standard SI prefixes.

Why should computer memory be the ONE EXCEPTION to the SI standard prefixes? We have binary prefixes. Use them if you want.

Re:SI units

[darthflo]

Everyone else except them uses KB/MB/TB/etc in a consistent - if not SI-compliant - fashion.

Bullshit. Remember 4.8 kbps modems which managed to transfer 4800 bps? What's the throughput of what's commonly referred to as Gigabit Ethernet, while we're on it? 1024 Mibps or more like 1'000'000'000 bps? What about an 1.5 Gbps SATA link? How many Pixels in a Megapixel? How many lomaniacs in a Megalomaniac?

Re:Seems Silly to me

[grahamwest]

You are right that it's actually 1440KB but you're totally wrong about everything else. 720KB floppies were double sided and "double density". That means 80 tracks, 18 sectors per track, 256 bytes per sector - with the 2 sides you get 720KB. "high density" or "quad density" floppies were 36 sectors per track, doubling them up to 1440KB. This is equivalent to 1.40625MB.

Re:wow.... are you clueless!

[name*censored*]

For starters, you could easily get an extra 24g of sugar (just as easily as you could get 24g less), whereas with hard drives you'll always get the same number of bytes in every identical model. Secondly, it'd be closer to buying a bag of sugar advertised as 1kg when it's really 2 pounds of sugar, or buying a car with an advertised fuel efficiency of 16 L/100mi when it's really 16 L/100km. Thirdly, computers work in base 2, so it makes sense to make disks in base 2 (unlike cars/explosives/sugar, which work in base 10).

I think it's about time that hard disk advertisers started claiming responsibility for what is perhaps the only unit disparity of it's kind (nothing else fudges the definition of a measurement unit to save a few bucks, as far as I'm aware). It's doubleplusgood that it's Seagate who's doing it, considering that they're leading the field and therefore will (hopefully) set an example to WD/Samsung/etc.

Re:SI units

[jez9999]

It's not a prefix, it's a suffix. £1000 involves a prefix, 1000MB involves a suffix.

This is Nonsense

[Lachlan Hunt]

The whole hard drive industry quotes storage capacities in base 10 SI units. Just because some ignorant consumers don't understand the difference between a Gigabyte and a Gibibyte doesn't mean that Seagate should have to pay for their ignorance. The customer got what he paid for. He should instead sue Operating System vendors for calculating storage capacities in base 2 and reporting as GB instead of either calculating in decimal or reporting as GiB.

Re:SI units

[Jah-Wren Ryel]

Only magnetic and optical storage have the luxury of defining units in non-power-of-2 ways, and yet they generally do not, choosing to standardize on 512-byte blocks primarily because if they didn't, the VM system's paging path would be heinously inefficient.

Using an OS's handling of RAM as a rationalization for 2^10 = K is a new one.

The fact is that NO modern hard disks have sector sizes of 512 bytes. You heard me, NONE. They commonly have payload sizes of 512 bytes, but the actual sector on a hard disk contains a lot more than just payload [freepatentsonline.com] - there are the ECC bits and the servo field which holds track, sector and disk head field grey code bits just to name the big ones. When added up, all the bits in a complete disk sector rarely equal a power of two, much less 2^10. Then are disks with 520 byte data payloads which are almost universally used in enterprise level disk arrays from manufacturers like HP, IBM, EMC, etc.

So, what's the point of that? Anyone who says disks naturally have power of 2 data organization as justification for saying 2^10 = 1K is just talking out of their ass.

...is sufficiently confusing to an average layman that it really doesn't work, either. Thus, the only -reasonable- choice is to standardize on base-2 definitions of these units.

Life is complicated, especially the technical parts. Unless you also propose to redefine data transmission rates like 100mbps ethernet and 150mbps SATA all your proposal does is rearrange the deck chairs.

Re:SI units

[redhog]

No, not RAM. Anything you _address_. You don't address the 412th cycle in the CPU frequence, you don't address the 1201st byte in transmission speed, etc. You address RAM content, and disk content and ports and hosts on the internet. All such addresses are stored as binary numbers inside the computer, and can thus address two to the power of number of bits in the address numbers of positions (bytes, hosts, bits, whatever).

Re:SI units

[Jah-Wren Ryel]

No, not RAM. Anything you _address_. You don't address the 412th cycle in the CPU frequence, you don't address the 1201st byte in transmission speed, etc. You address RAM content, and disk content and ports and hosts on the internet. All such addresses are stored as binary numbers inside the computer, and can thus address two to the power of number of bits in the address numbers of positions (bytes, hosts, bits, whatever).

Correction - you do not address disk content that way, you address filesystem content. Filesystems are defined by the host computer, disk sectors are not. Disk sectors exist independently from any host system and disks frequently contain an odd number of sectors, clearly not a power of two.

Furthermore, try finding a modern tape drive system with capacities measured in powers of two. You won't, they are all sized in base-10 units.

Re:RTFM

[joeljkp]

While he was a bit harsh, the 'gibi' 'mibi' 'mabi' and what not are NOT accepted units of measure irregardless of what any organization (even a 'standards' organization)says.

Binary prefixes are supported by the following professional organizations: IEEE, CIPM, SAE, NIST, CENELEC, and the European Union.

Source: http://en.wikipedia.org/wiki/Binary_prefix#Adoption [wikipedia.org]

Re:RTFM

[vux984]

Actually, most buses are rated in standard SI units, not base-2, this is because the "M" actually arrives by way of the clock speed rather than the storage size which is (often) bytes.

For example: consider AGP1x, which is rated at 266MB/s:

This is arrived at by multiplying: 66MHz clock x 4 byte data path (32 bits) = 266MB/s

(Note: actually 66 x 4 = 264, we get 266 due to round off error. The clock speed is more like to 66.6MHz than actually 66MHz. And 266MB/s is really just twice as fast as 133MB/s, which is the speed rating of regular PCI.)

The point however is that the M comes from the *M*Hz, and Hz are measured in base-10. 66.6MHz is 66,600,000 Hz

66,600,000 cycles per second x 4 bytes per cyle = 266,400,000 or rounded 266MB/s

That said, I agree with you. Storage measured in bytes is, by defacto standard, measured in base-2, where M=2^20, not 10^3.