WEBVTT

00:00:00.000 --> 00:00:02.000
So welcome to the deep dive. We're going to try

00:00:02.000 --> 00:00:05.240
a little thought experiment to start us off today.

00:00:05.679 --> 00:00:08.900
I want you to imagine it's the early 1990s. A

00:00:08.900 --> 00:00:11.000
very different computing landscape back then.

00:00:11.099 --> 00:00:13.080
Oh, completely. You've got your computer, and

00:00:13.080 --> 00:00:15.699
the hard drive is, I mean, it's a mere 20 to

00:00:15.699 --> 00:00:19.980
80 megabytes, which is nothing today. And expanding

00:00:19.980 --> 00:00:24.239
it costs a staggering $10 per megabyte. Yeah,

00:00:24.239 --> 00:00:27.079
which was a huge barrier for, well, Pretty much

00:00:27.079 --> 00:00:29.739
everyone. Exactly. So what do you do? Do you

00:00:29.739 --> 00:00:32.920
empty your wallet for a new drive or do you turn

00:00:32.920 --> 00:00:36.359
to this almost magical software wizardry from

00:00:36.359 --> 00:00:39.060
the 90s known as disk compression? Because that's

00:00:39.060 --> 00:00:41.299
our mission today. We are exploring this technology

00:00:41.299 --> 00:00:44.320
that seemingly just double a computer's physical

00:00:44.320 --> 00:00:46.460
storage out of thin air. It really did feel like

00:00:46.460 --> 00:00:48.500
magic to users back then. I mean, you have to

00:00:48.500 --> 00:00:50.280
remember software was rapidly getting bigger

00:00:50.280 --> 00:00:52.920
and physical storage was incredibly expensive.

00:00:53.119 --> 00:00:55.880
Right. OK, let's unpack this. Because the idea

00:00:55.880 --> 00:00:58.000
of just downloading space sounds impossible.

00:00:58.100 --> 00:01:01.399
I always picture it like a magical suitcase.

00:01:01.539 --> 00:01:04.579
You have the suitcase, and the exact second you

00:01:04.579 --> 00:01:07.400
pack your clothes into it, it automatically shrinks

00:01:07.400 --> 00:01:09.219
them down to half their size. That's a great

00:01:09.219 --> 00:01:11.560
way to visualize it. Right. And then the second

00:01:11.560 --> 00:01:13.739
you pull a shirt out, it instantly expands back

00:01:13.739 --> 00:01:16.519
to its normal size. And the wild part is you

00:01:16.519 --> 00:01:18.659
never even have to press a compress button. It

00:01:18.659 --> 00:01:21.500
just happens. And that automatic nature is what

00:01:21.500 --> 00:01:23.790
grounds your analogy. in the actual technical

00:01:23.790 --> 00:01:26.549
reality of our sources, because we really need

00:01:26.549 --> 00:01:29.569
to establish the difference between standard

00:01:29.569 --> 00:01:32.950
file compression and true disk compression. Which

00:01:32.950 --> 00:01:35.230
are two very different beasts. Right, completely.

00:01:35.489 --> 00:01:37.849
With file compression, like zipping a file, it's

00:01:37.849 --> 00:01:39.890
interactive. You manually tell the system to

00:01:39.890 --> 00:01:42.189
pack it up, and you manually tell to extract.

00:01:42.810 --> 00:01:45.230
But disk compression was totally non -interactive.

00:01:45.290 --> 00:01:47.549
It operated entirely in the background. It was

00:01:47.549 --> 00:01:50.040
on the fly, completely transparent. and happened

00:01:50.040 --> 00:01:52.760
in real time. So the user, and I guess the software

00:01:52.760 --> 00:01:55.239
they were running, had no idea this was even

00:01:55.239 --> 00:01:57.739
happening. Exactly. Your word processor, your

00:01:57.739 --> 00:02:00.099
games, they didn't know they were asking for

00:02:00.099 --> 00:02:02.480
compressed data. They just asked the system for

00:02:02.480 --> 00:02:04.640
a file. And the disk compression utility would

00:02:04.640 --> 00:02:07.299
grab it, decompress it instantly, and hand it

00:02:07.299 --> 00:02:10.930
over. OK, so how did it actually pull off? this

00:02:10.930 --> 00:02:13.550
illusion because you're basically creating physical

00:02:13.550 --> 00:02:16.810
space using only software. Well, it did this

00:02:16.810 --> 00:02:19.610
by essentially hijacking the file access layer

00:02:19.610 --> 00:02:22.229
of the operating system. When you installed one

00:02:22.229 --> 00:02:25.110
of these utilities, instead of saving thousands

00:02:25.110 --> 00:02:28.009
of individual files scattered all over the physical

00:02:28.009 --> 00:02:30.069
magnetic platters of your hard drive. Which is

00:02:30.069 --> 00:02:32.509
how it normally works. Right. Instead of that,

00:02:32.830 --> 00:02:35.610
the software created a single massive hidden

00:02:35.610 --> 00:02:38.180
file on the host drive. They usually called it

00:02:38.180 --> 00:02:40.919
a virtual drive. So every single thing you saved

00:02:40.919 --> 00:02:43.479
was secretly being compressed and stuffed directly

00:02:43.479 --> 00:02:46.580
into this one giant monolithic file by a special

00:02:46.580 --> 00:02:48.819
device driver. Wait, if everything is living

00:02:48.819 --> 00:02:51.719
inside one giant hidden file, doesn't that mean

00:02:51.719 --> 00:02:54.080
the computer is basically lying to the user about

00:02:54.080 --> 00:02:56.620
what a drive even is? What's fascinating here

00:02:56.620 --> 00:02:59.219
is that yes, the computer was entirely abstracting

00:02:59.219 --> 00:03:01.810
reality. It was a complete deception. Yeah, the

00:03:01.810 --> 00:03:04.729
setup process was wild. The utility would first

00:03:04.729 --> 00:03:08.389
create this empty virtual drive, which again,

00:03:08.569 --> 00:03:11.949
is just one massive hidden file. And then it

00:03:11.949 --> 00:03:14.330
would systematically transfer almost all of your

00:03:14.330 --> 00:03:16.789
files from the physical host drive into that

00:03:16.789 --> 00:03:18.629
compressed file. Wait, you said almost all. Did

00:03:18.629 --> 00:03:21.370
it leave some stuff behind? It had to. You couldn't

00:03:21.370 --> 00:03:24.199
compress vital system files. specifically the

00:03:24.199 --> 00:03:28.120
swap file. Because if you compress the swap file,

00:03:28.360 --> 00:03:30.759
which manages virtual memory, the computer would

00:03:30.759 --> 00:03:33.159
basically lock up trying to decompress its own

00:03:33.159 --> 00:03:35.580
working memory to function. So it leaves that

00:03:35.580 --> 00:03:37.120
alone, moves everything else into the hidden

00:03:37.120 --> 00:03:39.479
file, and then it simply inflates the reported

00:03:39.479 --> 00:03:41.300
size of that virtual drive. So it just tells

00:03:41.300 --> 00:03:43.020
the computer, hey, look at all this free space

00:03:43.020 --> 00:03:45.460
I suddenly have. Exactly. It reports the expanded

00:03:45.460 --> 00:03:48.199
theoretical size, leaving the operating system

00:03:49.060 --> 00:03:51.379
totally oblivious to the real physical limits.

00:03:51.580 --> 00:03:54.439
But, I mean, if the entire C drive is secretly

00:03:54.439 --> 00:03:57.060
just a hidden file on the physical drive, how

00:03:57.060 --> 00:03:59.419
does the computer even wake up? Like, when you

00:03:59.419 --> 00:04:01.979
turn a PC on, it needs to read the C drive to

00:04:01.979 --> 00:04:04.379
boot, but if the drive is fake and the driver

00:04:04.379 --> 00:04:06.860
needed to read it is locked inside the fake drive,

00:04:06.879 --> 00:04:09.020
that's... Well, it's a total chicken and egg

00:04:09.020 --> 00:04:11.840
problem. It is a massive paradox. And solving

00:04:11.840 --> 00:04:14.000
it required some incredible sleight of hand during

00:04:14.000 --> 00:04:16.420
the boot sequence. Let's look at the historical

00:04:16.420 --> 00:04:19.120
DOS boot process from the sources to see how

00:04:19.120 --> 00:04:21.759
they trick the system. OK, lay it on me. So you

00:04:21.759 --> 00:04:24.259
press the power button. The system BIOS wakes

00:04:24.259 --> 00:04:26.360
up and it looks at the very first sector of the

00:04:26.360 --> 00:04:29.319
hard drive, the master boot record, or MBR. Right,

00:04:29.480 --> 00:04:31.920
the absolute baseline hardware level. The MBR

00:04:31.920 --> 00:04:34.910
then loads the volume boot record. which in turn

00:04:34.910 --> 00:04:38.209
loads the DOS BIOS. These are hidden system files,

00:04:38.529 --> 00:04:43.069
usually iOSYS or IBMBIO .com. At this stage,

00:04:43.269 --> 00:04:45.769
DOS is loading into basic memory, but, and this

00:04:45.769 --> 00:04:48.009
is the crucial part, it hasn't loaded the config

00:04:48.009 --> 00:04:50.730
-dos -sys file yet. Meaning no third -party drivers

00:04:50.730 --> 00:04:52.790
are loaded. The computer still thinks the physical

00:04:52.790 --> 00:04:55.370
hardware is the real C -DRAF. Exactly. And if

00:04:55.370 --> 00:04:56.970
it kept booting normally, it would just crash,

00:04:57.230 --> 00:04:59.310
because all the user files and the rest of the

00:04:59.310 --> 00:05:02.470
OS are trapped inside that unreadable compressed

00:05:02.470 --> 00:05:05.629
monolith. So how did they bypass the crash? In

00:05:05.629 --> 00:05:09.230
MS -DOS 6 .0, Microsoft introduced an undocumented

00:05:09.230 --> 00:05:12.009
challenge response interface known as the Preload

00:05:12.009 --> 00:05:14.790
API. Here's where it gets really interesting,

00:05:14.949 --> 00:05:17.970
because I picture this like intercepting the

00:05:17.970 --> 00:05:20.089
computer's morning coffee. The machine is just

00:05:20.089 --> 00:05:22.170
waking up going through its groggy morning routine,

00:05:22.449 --> 00:05:25.009
and this Preload API just steps in front of it.

00:05:25.129 --> 00:05:26.970
That's a perfect way to put it. Right after the

00:05:26.970 --> 00:05:30.089
DOS BIOS loads, but literally milliseconds before

00:05:30.089 --> 00:05:32.790
it reads the user's config files, the system

00:05:32.790 --> 00:05:36.370
pauses. It scans the root directory for specific

00:05:36.370 --> 00:05:39.490
preload drivers, files like dblspace .bin or

00:05:39.490 --> 00:05:41.730
stacker .bin. So compression drivers? Right.

00:05:41.790 --> 00:05:44.170
And if it finds one, it initiates this secret

00:05:44.170 --> 00:05:47.069
handshake. If the driver gives the correct mathematical

00:05:47.069 --> 00:05:50.370
response, the system loads it into memory immediately,

00:05:50.870 --> 00:05:53.110
completely bypassing normal configuration. And

00:05:53.110 --> 00:05:55.870
what does the driver do once it's in? It immediately

00:05:55.870 --> 00:05:58.620
executes a radical maneuver. It literally swaps

00:05:58.620 --> 00:06:01.639
the drive letters in the system's RAM. It swaps

00:06:01.639 --> 00:06:03.939
them. So it puts on a mask and says, hey, look

00:06:03.939 --> 00:06:07.079
at me. I am the C drive now. Yes. The physical

00:06:07.079 --> 00:06:09.620
drive is shoved aside, usually reassigned to

00:06:09.620 --> 00:06:12.399
a random letter like H, and the fake compressed

00:06:12.399 --> 00:06:15.269
file is mounted and declared the C drive. And

00:06:15.269 --> 00:06:17.750
from that millisecond on, the computer finishes

00:06:17.750 --> 00:06:20.410
its morning coffee, completely oblivious that

00:06:20.410 --> 00:06:22.889
its entire reality was just remapped. That is

00:06:22.889 --> 00:06:25.930
just wild. But I mean, tricking the computer's

00:06:25.930 --> 00:06:28.009
brain and stuffing literally all your data into

00:06:28.009 --> 00:06:31.490
a single hidden file sounds incredibly risky.

00:06:31.589 --> 00:06:33.790
Like, what were the unintended consequences of

00:06:33.790 --> 00:06:35.430
doing this? Because this has to have a catch.

00:06:35.550 --> 00:06:37.509
Oh, there were several massive catches. But before

00:06:37.509 --> 00:06:39.689
we even get to the data disasters, we have to

00:06:39.689 --> 00:06:42.329
talk about the performance paradox. OK. You'd

00:06:42.329 --> 00:06:44.509
naturally assume that forcing a computer to do

00:06:44.509 --> 00:06:47.589
complex math to compress and decompress every

00:06:47.589 --> 00:06:50.009
single file in real time would make the whole

00:06:50.009 --> 00:06:51.509
system sluggish. Right, yeah. I mean, you're

00:06:51.509 --> 00:06:54.930
giving the CPU a massive extra chore for every

00:06:54.930 --> 00:06:58.319
single task. The sources show it actually depended

00:06:58.319 --> 00:07:00.899
on where your bottleneck was. If your system

00:07:00.899 --> 00:07:03.620
was CPU -bound, meaning the processor was already

00:07:03.620 --> 00:07:06.420
struggling to keep up with basic tasks, then

00:07:06.420 --> 00:07:08.680
yes, disk compression slowed everything to a

00:07:08.680 --> 00:07:10.980
crawl. The math was just too heavy. So it was

00:07:10.980 --> 00:07:13.560
just choking the processor? Right. But if your

00:07:13.560 --> 00:07:15.339
system was IO -bound, meaning you had a fast

00:07:15.339 --> 00:07:18.439
processor, but a really slow physical hard drive.

00:07:18.480 --> 00:07:20.439
Which was pretty common back then, waiting for

00:07:20.439 --> 00:07:23.000
the physical platters to spin. Exactly. For those

00:07:23.000 --> 00:07:25.800
IO -bound machines, disk compression actually

00:07:25.800 --> 00:07:28.279
increased overall performance. Wait, really?

00:07:28.500 --> 00:07:31.180
It made the computer faster? How? Because there

00:07:31.180 --> 00:07:33.420
was physically less data to read and write from

00:07:33.420 --> 00:07:36.279
the slow mechanical drive, the fast processor

00:07:36.279 --> 00:07:39.040
could read a tiny compressed file from the disk

00:07:39.040 --> 00:07:42.079
and decompress it in RAM significantly faster

00:07:42.079 --> 00:07:43.939
than the slow hard drive needle could read a

00:07:43.939 --> 00:07:46.560
massive uncompressed file. Oh, wow. So the math

00:07:46.560 --> 00:07:48.560
was actually faster than the physics of the spinning

00:07:48.560 --> 00:07:51.100
disk. That's brilliant. But obviously, performance

00:07:51.100 --> 00:07:54.579
aside, the risks to your actual data had to be

00:07:54.579 --> 00:07:57.579
astronomical. They were. The nightmare scenarios

00:07:57.579 --> 00:08:00.240
usually began with a very simple user action,

00:08:00.939 --> 00:08:03.220
trying to uninstall the software. Because it's

00:08:03.220 --> 00:08:06.930
a trap. A total trap. If your expanded uncompressed

00:08:06.930 --> 00:08:09.769
data exceeded the physical size of the actual

00:08:09.769 --> 00:08:12.529
hard drive, uninstallation was mathematically

00:08:12.529 --> 00:08:15.370
impossible unless you deleted enough files to

00:08:15.370 --> 00:08:17.629
fit back into reality. You literally couldn't

00:08:17.629 --> 00:08:19.990
escape the virtual space you built. Exactly.

00:08:20.649 --> 00:08:23.689
And it gets worse. A massive danger lurk during

00:08:23.689 --> 00:08:26.620
the initial setup phase. The compression utility

00:08:26.620 --> 00:08:29.439
relied completely on the integrity of your drive's

00:08:29.439 --> 00:08:32.059
FAT file system to move your data safely. The

00:08:32.059 --> 00:08:34.519
file allocation table, basically the index card

00:08:34.519 --> 00:08:36.740
catalog for where data lives on the disk. Right.

00:08:36.960 --> 00:08:39.659
So if your drive had pre -existing errors, particularly

00:08:39.659 --> 00:08:42.019
something called cross -link files, where two

00:08:42.019 --> 00:08:44.399
files accidentally claim the same physical cluster

00:08:44.399 --> 00:08:46.860
on the disk, the results were catastrophic. Uh

00:08:46.860 --> 00:08:48.860
-oh. Because the setup program is sweeping through,

00:08:49.019 --> 00:08:51.259
grabbing clusters, compressing them, and then

00:08:51.259 --> 00:08:53.960
what? Deleting the originals to save space? Yes.

00:08:54.190 --> 00:08:56.629
It would compress the cluster for the first file,

00:08:56.870 --> 00:08:59.649
delete the original, and move on. So when it

00:08:59.649 --> 00:09:01.129
eventually got to the second file that claimed

00:09:01.129 --> 00:09:04.970
that exact same physical space, the data was

00:09:04.970 --> 00:09:08.309
already gone. Oh, man. Massive data loss during

00:09:08.309 --> 00:09:10.149
the initial transaction. Massive. It would corrupt

00:09:10.149 --> 00:09:12.149
files and could even destroy the virtual translation

00:09:12.149 --> 00:09:14.750
table itself, ruining the entire drive before

00:09:14.750 --> 00:09:17.009
the installation even finished. Which leads to

00:09:17.009 --> 00:09:20.169
my biggest question. What happened if a user

00:09:20.169 --> 00:09:22.399
was just, you know, poking around their hard

00:09:22.399 --> 00:09:24.700
drive, making hidden files visible, and they

00:09:24.700 --> 00:09:27.720
saw the suspiciously massive file eating up basically

00:09:27.720 --> 00:09:30.639
100 % of their space, what if they just deleted

00:09:30.639 --> 00:09:32.980
it? I mean, that is the most human error of all.

00:09:32.980 --> 00:09:35.480
Yeah. Because the file was hidden, but curious

00:09:35.480 --> 00:09:38.100
users often found it. And because they didn't

00:09:38.100 --> 00:09:40.440
grasp the abstraction, they didn't realize that

00:09:40.440 --> 00:09:42.740
one file was their entire operating system, and

00:09:42.740 --> 00:09:44.940
all their personal data, they'd hit Delete to

00:09:44.940 --> 00:09:47.600
free up space. Oh. Total annihilation. Deleting

00:09:47.600 --> 00:09:49.840
that one file meant losing every single thing

00:09:49.840 --> 00:09:52.299
stored on the compressed drive instantly. Just

00:09:52.299 --> 00:09:55.580
boom, your whole digital life is gone. But despite

00:09:55.580 --> 00:09:59.299
that risk, people still used it because, well,

00:09:59.480 --> 00:10:01.899
it was either this or shell out hundreds of dollars

00:10:01.899 --> 00:10:04.879
for a new drive. Exactly. The high cost of hard

00:10:04.879 --> 00:10:07.299
drives made this software an absolute necessity,

00:10:07.299 --> 00:10:10.019
which sparked a massive multi -million dollar

00:10:10.019 --> 00:10:12.379
industry arms race. So let's talk about that

00:10:12.379 --> 00:10:15.259
arms race because the tech really had to evolve

00:10:15.259 --> 00:10:17.710
to keep up, right? It did. In the early days,

00:10:17.970 --> 00:10:20.490
CPUs were just too weak to handle the math. So

00:10:20.490 --> 00:10:22.470
the first solutions were actually hardware -assisted.

00:10:22.590 --> 00:10:24.789
Oh, like physical add -on boards? Yeah. Companies

00:10:24.789 --> 00:10:27.230
like Stack Electronics with their stacker product,

00:10:27.610 --> 00:10:31.029
or InfoChip Systems with Expanse, they sold physical

00:10:31.029 --> 00:10:33.330
coprocessor boards that plugged into the motherboard.

00:10:33.970 --> 00:10:36.490
These had custom chips designed specifically

00:10:36.490 --> 00:10:38.690
to do the compression math, taking the load off

00:10:38.690 --> 00:10:41.850
the main CPU. But eventually processors got faster,

00:10:41.889 --> 00:10:44.509
like the 486 and the Pentium? Right. And once

00:10:44.509 --> 00:10:47.080
the CPUs gained enough power, the hardware boards

00:10:47.080 --> 00:10:50.240
became obsolete. Software -only solutions completely

00:10:50.240 --> 00:10:53.399
took over the market. We saw programs like Squish,

00:10:53.779 --> 00:10:56.879
Superstore, and DoubleDisk dominating. And computer

00:10:56.879 --> 00:10:58.820
resellers loved this, right? I mean, they were

00:10:58.820 --> 00:11:01.200
bundling this software onto PCs just to claim

00:11:01.200 --> 00:11:03.059
their computers had more storage without actually

00:11:03.059 --> 00:11:04.919
buying bigger hard drives. Oh, it was a massive

00:11:04.919 --> 00:11:07.340
marketing gimmick. And the operating system developers

00:11:07.340 --> 00:11:10.700
caught on quickly. Digital research bundled Superstore

00:11:10.700 --> 00:11:14.240
into DockerDOS 6 .0, and Microsoft bundled DoubleSpace,

00:11:14.679 --> 00:11:17.500
which licensed DoubleDisk Tech into MS -DOS 6

00:11:17.500 --> 00:11:19.580
.0. Okay, I wanted to ask you about this because

00:11:19.580 --> 00:11:21.740
there's a very curious detail in the sources.

00:11:22.120 --> 00:11:25.559
I see that MS -DOS 6 .22 bundled a program called

00:11:25.559 --> 00:11:28.039
DriveSpace, but the version right before it,

00:11:28.100 --> 00:11:31.179
MS -DOS 6 .21, shipped it with no compression

00:11:31.179 --> 00:11:34.120
software at all. And the sources mention legal

00:11:34.120 --> 00:11:36.580
reasons. What does that tell us about the climate

00:11:36.580 --> 00:11:39.330
back then? Impartially speaking, without getting

00:11:39.330 --> 00:11:41.649
into the weeds of the specific patent lawsuits,

00:11:42.330 --> 00:11:44.769
that gap in MS -DOS versions really highlights

00:11:44.769 --> 00:11:47.330
how intensely competitive and legally fraught

00:11:47.330 --> 00:11:49.350
this industry had become. I mean, it was virtual

00:11:49.350 --> 00:11:52.159
real estate. Exactly. Companies were aggressively

00:11:52.159 --> 00:11:54.639
patenting their algorithms and their integration

00:11:54.639 --> 00:11:57.299
techniques. It was a highly lucrative market.

00:11:57.679 --> 00:12:00.700
And the fact that a flagship OS like MS -DOS

00:12:00.700 --> 00:12:03.240
had to temporarily strip out its biggest feature

00:12:03.240 --> 00:12:06.500
shows you just how fierce the legal battles over

00:12:06.500 --> 00:12:08.399
this intellectual property were. Right. Everybody

00:12:08.399 --> 00:12:10.769
wanted to own the rights to thin air. And the

00:12:10.769 --> 00:12:13.389
technology was getting incredibly advanced before

00:12:13.389 --> 00:12:15.950
the whole industry eventually collapsed. For

00:12:15.950 --> 00:12:19.009
instance, the sources detail how Novell DOS 7

00:12:19.009 --> 00:12:22.389
bundled a DPMS -enabled version of Stacker. What

00:12:22.389 --> 00:12:24.909
does that mean for the user? It meant that if

00:12:24.909 --> 00:12:27.610
you had Stacker on your local machine and Stacker

00:12:27.610 --> 00:12:30.370
on a remote network server, they could exchange

00:12:30.370 --> 00:12:32.830
data packets across the network while keeping

00:12:32.830 --> 00:12:35.200
them compressed. Wait, so they didn't have to

00:12:35.200 --> 00:12:37.879
uncompress the file, send it, and recompress

00:12:37.879 --> 00:12:40.460
it? They just sent the shrunken version. Exactly.

00:12:40.779 --> 00:12:42.700
Which effectively doubled the network bandwidth.

00:12:42.879 --> 00:12:45.019
It was a brilliant use of the technology to speed

00:12:45.019 --> 00:12:47.700
up entire local area networks. That's actually

00:12:47.700 --> 00:12:50.740
amazing. So what does this all mean for the listener

00:12:50.740 --> 00:12:54.080
today? I mean, where did all this fiercely contested

00:12:54.080 --> 00:12:57.980
tech actually go? Because we don't buy box software

00:12:57.980 --> 00:13:00.639
to double our drives anymore. It essentially

00:13:00.639 --> 00:13:03.970
faded into disuse by the late 1990s. Hard drive

00:13:03.970 --> 00:13:07.230
manufacturing advanced so rapidly and the cost

00:13:07.230 --> 00:13:10.350
per megabyte dropped so low that physical storage

00:13:10.350 --> 00:13:12.990
became incredibly cheap. So it just didn't make

00:13:12.990 --> 00:13:15.429
sense to risk your data with a virtual drive

00:13:15.429 --> 00:13:18.450
anymore. Right. But the DNA of the technology

00:13:18.450 --> 00:13:22.149
absolutely survives today. Windows still supports

00:13:22.149 --> 00:13:25.350
NTFS compression, which blends traditional file

00:13:25.350 --> 00:13:27.950
compression with those transparent on the fly

00:13:27.950 --> 00:13:29.649
concepts we just talked about. So it's still

00:13:29.649 --> 00:13:31.820
running in the background. Yeah. And the open

00:13:31.820 --> 00:13:34.279
source community kept the legacy alive too. LENX

00:13:34.279 --> 00:13:37.460
developers created drivers like DMSDS specifically

00:13:37.460 --> 00:13:40.080
to read those old 90s stacker and double space

00:13:40.080 --> 00:13:42.659
volumes preserving that history. Which is pretty

00:13:42.659 --> 00:13:45.360
cool. It just quietly became part of the background

00:13:45.360 --> 00:13:47.879
architecture of how we compute. Which actually

00:13:47.879 --> 00:13:49.919
leaves us with a really fascinating final thought

00:13:49.919 --> 00:13:52.279
to mull over today. Yeah, because for an entire

00:13:52.279 --> 00:13:54.580
generation of computer users, disk compression

00:13:54.580 --> 00:13:57.080
was their very first introduction to the concept

00:13:57.080 --> 00:13:59.960
of virtualization. It was the first time we learned

00:13:59.960 --> 00:14:01.879
that the hardware you think you're interacting

00:14:01.879 --> 00:14:04.399
with, the space, the folders, the drive, can

00:14:04.399 --> 00:14:07.240
actually just be a software illusion. That's

00:14:07.240 --> 00:14:08.960
a really good point. It completely decoupled

00:14:08.960 --> 00:14:11.460
the physical reality from the digital experience.

00:14:11.879 --> 00:14:15.840
Exactly. So, as you go about your day today,

00:14:16.360 --> 00:14:18.759
you look at your cloud storage or use virtual

00:14:18.759 --> 00:14:22.279
machines or log into simulated environments,

00:14:22.759 --> 00:14:25.480
ask yourself, how much of your current digital

00:14:25.480 --> 00:14:28.559
world is actually real physical hardware? And

00:14:28.559 --> 00:14:30.940
how much is just a modern version of a giant

00:14:30.940 --> 00:14:33.440
hidden file pretending to be a hard drive? Something

00:14:33.440 --> 00:14:35.659
to think about. Definitely. We'll leave you with

00:14:35.659 --> 00:14:37.539
that. Thanks for joining us on this deep dive.
