As the article belatedly mentions, the ~40 GB automated tape library for HARVEST was the bee's knees... and possibly the inspiration for "harvest now, encrypt later".
PS – Primary sources for the IBM 7950/HARVEST on Bitsavers, excellent as always:
Every time I read an article like this, I get depressed about all the hours of my life I've spent waiting for computers. I'm 64 now, and I've been using computers since I was 14. (Batch mode, using punched cards on an IBM 370 was the first computer I programmed, while I was in high school.)
One of the big features of the Harvest was the so called POPCOUNT which was supposedly added the machine at the NSA's request. (This is the number of bits in a word that are on.) I've always wondered how this might be used. Some suggest computing the Hamming distance. Others suggest some kind of statistical test like chi-squared. But are there other more algebraic solutions that can benefit from it?
They had vertical integration and made bank. Then each layer got more complex as computing became mainstream, and IBM got out-competed on everyone of them by specialized companies. And then they sold them off. The only layer that remains is the most strategic: the customer relationship.
In some areas IBM still holds if not a technological advantage at least the first among equals position, they just aren't areas that are flashy or generally known outside of folks like us - Mainframes and related technologies for example, the IBM Z series is still a huge player there.
Sometime in the 90s they decided that becoming a services and consultancy provider was more important than being involved in large-scale commodity hardware or really anything else.
Over three decades they've exited commodity hardware (x-series and thinkpads), retreated from a lot of software stuff and generally gone as hard as they can for large-scale business contracts. They seem to keep a decent amount of cutting-edge hardware research going, Power is still broadly competitive with x86 AFAICT, but the volumes have got to be pretty low these days and I half suspect that some of it is just a "Hey look at us we're still here" play rather than actually feeding into products, because services are the profit-centre.
They're still huge and have a lot of money, and they own Redhat. But IMHO it's in a long, slow decline, and it has stepped back from the public eye and most public-facing markets.
Context - I've worked for them in a few places. Obvs my opinons are my own, especially as I've been 'out' for over a decade at this point.
They royally screwed up by signing an exclusive deal with the back-then no-name Microsoft company. IBM created the 3 trillion behemoth that Microsoft is while killing themselves in the process.
It's really sad but also infuriating: had IBM not fucked up and not signed Microsoft, that mediocre company that Microsoft is would have gone nowhere and we wouldn't have to suffer their turds since decades.
P.S: in their dying breath (and I know IBM still exists), they did commit one billion dollar to Linux early on (which back then was a big deal). When we see the tens of billions of devices that now run Linux, IBM at least partially participated in Linux taking over the world. So there's that.
As the article belatedly mentions, the ~40 GB automated tape library for HARVEST was the bee's knees... and possibly the inspiration for "harvest now, encrypt later".
PS – Primary sources for the IBM 7950/HARVEST on Bitsavers, excellent as always:
https://www.bitsavers.org/pdf/ibm/7950/
Decrypt later, I suppose.
Ouch, that was indeed a typo...
Every time I read an article like this, I get depressed about all the hours of my life I've spent waiting for computers. I'm 64 now, and I've been using computers since I was 14. (Batch mode, using punched cards on an IBM 370 was the first computer I programmed, while I was in high school.)
One of the big features of the Harvest was the so called POPCOUNT which was supposedly added the machine at the NSA's request. (This is the number of bits in a word that are on.) I've always wondered how this might be used. Some suggest computing the Hamming distance. Others suggest some kind of statistical test like chi-squared. But are there other more algebraic solutions that can benefit from it?
Has anything ever been written up about the chips that the was (is) NSA fabbing?
Wow... so many blinkenlights! Love it.
"IBM's Fred Brooks Jr. (..)"
Yes, that guy from "The Mythical Man-Month" fame.
Cold War era computing is fascinating. The scale they achieved with 1960s technology is impressive.
You'll want to read about SAGE
en.wikipedia.org/wiki/Semi-Automatic_Ground_Environment
Also Western Electric's CLC for Safeguard.
sometimes I wish I was born a few decades earlier. it would have been fun to work on things like this when they were new
What happened to IBM?
They had vertical integration and made bank. Then each layer got more complex as computing became mainstream, and IBM got out-competed on everyone of them by specialized companies. And then they sold them off. The only layer that remains is the most strategic: the customer relationship.
IBM is now just a consultancy.
Not much is left of the great IBM, but their Z architecture is still superior when it comes to reliability at scale.
In fact, "what happened to IBM?", they were in the news recently: https://wccftech.com/ibm-worlds-first-dual-architecture-proc...
Creating the world's first dual instruction set CPU.
> Creating the world's first dual instruction set CPU.
World's first? What about the VIA C3 (or earlier NEC V20)? I'm sure there are more / earlier ones.
https://en.wikipedia.org/wiki/Alternate_Instruction_Set
That's not the same thing, it's more like a hidden feature than the ISA of an entirely different CPU.
There was also the semi-mythical PowerPC 615 that was meant to have both PowerPC and x86 instruction sets. W
I think that's a little simplistic/reductive.
In some areas IBM still holds if not a technological advantage at least the first among equals position, they just aren't areas that are flashy or generally known outside of folks like us - Mainframes and related technologies for example, the IBM Z series is still a huge player there.
They are also a consultancy though.
They still produce the mainframe hardware ranges though.
> IBM is now just a consultancy.
Except that they own Red Hat
Sometime in the 90s they decided that becoming a services and consultancy provider was more important than being involved in large-scale commodity hardware or really anything else.
Over three decades they've exited commodity hardware (x-series and thinkpads), retreated from a lot of software stuff and generally gone as hard as they can for large-scale business contracts. They seem to keep a decent amount of cutting-edge hardware research going, Power is still broadly competitive with x86 AFAICT, but the volumes have got to be pretty low these days and I half suspect that some of it is just a "Hey look at us we're still here" play rather than actually feeding into products, because services are the profit-centre.
They're still huge and have a lot of money, and they own Redhat. But IMHO it's in a long, slow decline, and it has stepped back from the public eye and most public-facing markets.
Context - I've worked for them in a few places. Obvs my opinons are my own, especially as I've been 'out' for over a decade at this point.
They royally screwed up by signing an exclusive deal with the back-then no-name Microsoft company. IBM created the 3 trillion behemoth that Microsoft is while killing themselves in the process.
It's really sad but also infuriating: had IBM not fucked up and not signed Microsoft, that mediocre company that Microsoft is would have gone nowhere and we wouldn't have to suffer their turds since decades.
P.S: in their dying breath (and I know IBM still exists), they did commit one billion dollar to Linux early on (which back then was a big deal). When we see the tens of billions of devices that now run Linux, IBM at least partially participated in Linux taking over the world. So there's that.
They became NSC, no such company.
That one job was to process "sensitive" intelligence intercepts and break codes, if you're not grabbed by the clickbait headline. It's a good read.
What's clickbaity about it? Breaking codes and using that to process intercepts (or other encrypted text) is the definition of cryptanalysis.
[dead]