Hacker News new | comments | show | ask | jobs | submit login
How Not to Network a Nation: The Uneasy History of the Soviet Internet (uic.edu)
116 points by ingve on May 2, 2016 | hide | past | web | favorite | 59 comments

One more author rant before the thread vanishes, here are a couple points from the book.

1. The Soviet internet is a sideways allegory for the present. The Soviet Union, once we work through its parade of horribles, helps separate us from and then rethink our current network values. General secretaries, whether state or corporate, have long been trying to privatize our information.

2. The reasons the Soviets did not develop a network are not the reasons we often like to think: it's not because networks are anathema to censorship and control structures (think cybersecurity and dictatorships today), it's not because of technological backwardness (que Soviet military networks since the mid 1950s), it's not exactly because their genuinely screwed up command economy was either too rigid or hierarchical (que the rest of the book).

3. The Soviet story is a tragedy of big science and state support gone wrong, as well as a cautionary tale for how we go about building the network future in those terms.

And a link dump: enjoy!

Review by David Strom


Review by Michael Gordin in Nature


Podcast segment with Kerri Smith (starts at 6:50) in Nature


Conversation in The Atlantic.


So, I'm getting the sense this is related to Glushkov's work on pricing right? My understanding, please correct me if I am wrong, was to try and track all items and use computer systems to optimize production and pricing. Was that integral to the network he envisioned? Do you touch much on the various economic debates in the Soviet Academy at the time on how to run a non-market system, like Kantorovich's ideas?

That's right. The network project was the means Glushkov envisioned for realizing both economic optimization and a series of other technical upgrades, both wild (like mind uploading) and mundane (like paperless office). Yes, Kantorovich, Glushkov, and the economic cybernetic regulation of market without the market (contra Kosygin-Liberman reforms) figure squarely into the central chapters.

how does this compare to an overlooked success of Soviet Tech ie CDMA..which was their invention..

Great lead. Glushkov's network ideas are definitely overlooked, but the parallel with the CDMA in weak. Perhaps a stronger parallel is with, say, Lashkaryov's p-n junction, which became an important innovation (semiconductor) but thanks to others elsewhere. Loren Graham's Lonely Ideas is a great book for more on the larger history of brilliant ideas and halting innovations.


Thanks, pjc50, for the helpful reference. This is the author of the book here. Your instincts are sharp. As it happens, I deal with exactly your last argument in the book, so let me gently suggest that if in fact there were "no way" for anything but central control, this book would not exist because there would be no story to tell. However, it does, and the three-decade Soviet story told here, whose protagonist is a leading theorized of in fact decentralized power, might deserve another glance...

The same thing happened in France, kind of.

For a while (in the 70s and 80s) there were two competing, state-sponsored networks in France: Minitel, based on x25 and made by French telcos, and Cyclades, based on datagrams and made by French CS research labs. The telcos lobbied the government to cut funding for Cyclades, which they saw as a competitor, and it eventually happened (under Valéry Giscard d'Estaing, probably neither Pompidou nor de Gaulle would have made that mistake).

There is a very good book (in French) about that story, called La France en réseaux, by Valérie Schafer. It is one of my favorite tech history books. I will certainly read yours and hope it is as good!

However, I saw it is only available as hardcover. Any chance you can make it available for Kindle?

Thanks, catwell. Super helpful. Schafer's work comes highly recommended. I'll get to it!

MIT Press put out an e-book version, and the soft cover comes out later this year. (A pdf is probably out there too...)

A number of African nations are currently struggling with this -- attempting to bootstrap nationwide networking from zero, while having economies which are very much command economies (although in Africa, that's due to corruption, rather than being baked into the law).

In one small African nation the government was intensely interested in building a national system for finding lost cows. (This was a much-touted presidential initiative.) Basically, farmers would be able to text their cow's ID and the system would tell them if another farmer had found it.

In spite of the simplicity of such a system, it took them years of 'research and development', and enormous budgets spent, for a system which they discovered in the end, didn't work. It turned out, they couldn't figure out how to procure and install a server for it to run on.

>It turned out, they couldn't figure out how to procure and install a server for it to run on.

Sure they could. There's talent abound in Africa. Unfortunately, the case here is usually corruption. That project money was siphoned to the politicians themselves or a connected pal, or both. Just old fashioned fraud here. I think "Africans don't know what they are doing" is a convenient scapegoat that empowers fraudsters. And if that narrative leads to more foreign aid, then wonderful, because the politically connected class abuse that too.

Oh, I'm glad you know more than I do about this thing I lived through, so you could correct me on it.

Western developed economies are quite capable of doing this, such as the Canadian gun registry which somehow managed to spend billions of dollars without achieving anything.

CGI, which built the registry, is the parent company of CGI Federal, known lately for the Obamacare site.

Let me guess... they're absurdly well connected though?

You don't get large federal contracts without being well-connected.

And this sub-thread is one, if tiny and partial, solution. Sunlight (public knowledge and understanding), bleaching out such stains. I appreciate the pointers.

And... we shouldn't let these people simply re-organize under new names.

Just as I'm likely to look askance at anyone with Goldman Sachs in their resume. Only, I prefer to know and respond more specifically.

And before that they were known for bungling a healthcare exchange for Canada's CMS (Centers for Medicare & Medicaid Services).

I guess the US took them because they had experience in the field?

Off topic of course. The gun registry was very political and had many people who wanted it to fail. Then it did fail. I always wondered if the two were related.

Then again, it could've failed simply because it was brought to you by the makers of healthcare.gov

(well, I guess the firearms registry was first, but you get the idea about the creators)

And despite this immense failure, in Québec we are looking to do it all over again.


Shouldn't Twillio, digital ocean and one small bespoke node app be able to do it?

It's not about what needs to be done, it's about finding reliable people to do it, especially when there's all sorts of opportunities for people to funnel it to a "preferred bidder".

If half of them would work outside of the US, people would already be happy. DO’s usability goes down a lot if you don’t have a credit card (you literally pay to be able to pay), and Twilio is quite similar.

The government was immensely paranoid about it, insisting it had to run on site, and be completely configurable by their tech staff, who were not themselves engineers.

See also "Project Cybersyn" for another attempt at making a computerised command economy work. It's referenced in the article.

While this looks like an fascinating read, it's worth remembering that the thing that makes the Internet the Internet, distinguishing it from other Western systems like X25, Minitel, and telecoms in general is its peer-to-peer nature with minimal central arbitration(+).

The Soviet Union rationed access to photocopiers. There was no way it would ever build a network that was not entirely subject to central control, and this must inevitably handicap its usage and development.

(+) Minimal, not zero. The central point for a long time was Jon Postel.

Project Cybersyn is interesting as a kind of alternate-history thought experiment, but imo it's not that useful as an example to study historically to learn empirical lessons from, because of how early it was canceled (due to military coup). It was first proposed in mid-1971, with design and prototyping starting in 1972, and Allende was overthrown in 1973, when it was still at a prototype stage. If it had run for a few more years, there would've been a lot more to go on.

Actually if you compare the cost and quality of internet service in US and some of the states formerly part of USSR it will make you very sad. An equivalent of comcast $50 plan will run you $5 and that's with no caps or shaping . Looking at less then stellar origins of internet in USSR you can see how monopolization f$%ed up US ISP market.

Former Soviet states have lower incomes, which means both a smaller ability to pay for services, as well as a lower cost of providing services (building and maintaining infrastructure, installation, support, those types of wages) on behalf of the ISP.

Other European states that were never Soviet also have much cheaper much faster average internet connections than the US... but these places are much smaller and more easily connected and maintained, I suppose.

Anyway, the picture isn't so black and white, maybe we get screwed in the US, but maybe we don't or not as badly as you suggest.

Tangentially, I am amazed you can get unlimited satellite cell internet for $125/mo from Iridium now, I'm sure that plan is slow as molasses but... its global and that's amazing to me.

Other European states that were never Soviet also have much cheaper much faster average internet connections than the US... but these places are much smaller and more easily connected and maintained, I suppose.

If this were truly the cause of the US's bandwidth backwardness, then denser pockets of the US should have been wired with gigabit access around the year 2000. Instead, urban areas often have much worse speeds and prices.

We get screwed in US. Install is one time cost equipment costs the same capital is 2-3x cheaper in US.

It's really the same story. Our command economy for bandwidth is run by NBC/Universal and whomever owns Charter.

From what I understand, in Soviet Russia/Eastern Bloc pretty much everything IT-related was copied from the West - various clones of IBM, DEC, Intel architectures. There were some original designs but they didn't really take off.

So given that I'm not really surprised 'soviet internet' didn't come to fruition even though I'm sure the book is an interesting read from the historical perspective.

This is a reasonable instinct, although, as I argue in the book, it really only explains things after the 1975 KGB-advised decision to harness the Soviet IT industries to cloning western innovations. Before that decision, between 1959 and 1975, when the Soviet networks are getting their start, the Soviet computing industry is often pioneering and interesting. After 1975, Soviet deliberate technological just-behindness does not equate to technological backwardness. One of the takeaways of this story is that peerless imaginative foresight, technological wizardry, and political prowess are not enough to change the world.

> between 1959 and 1975, when the Soviet networks are getting their start, the Soviet computing industry is often pioneering and interesting.

And when you really decide to see what was so pioneering in Soviet computing industry, it all boils down from 'pioneering' to only 'interesting'.

Which helps to get a perspective on a true scope of Soviet CS research.

After all, layered copying of Intel x86 CPUs also was 'pioneering and interesting'.

> After all, layered copying of Intel x86 CPUs also was 'pioneering and interesting'.

It was; at some periods I think Soviets were world leaders in electronics reverse engineering. Those techniques were rumored to get used well after USSR death in some Asian countries.

MESM -> BESM-6 machines I've heard had some abilities like a number of simultaneously supported IO ports (AS system for space program?) - which may or may not be considered pioneering. Another example is Setun ternary computer; of course that's providing we should consider some computers after ENIAC 'pioneering enough'.

That's the one that rings a faint bell, for me -- ternary systems. Although I've no background to speak of and suspect, based on their work in mathematics et al., that there is considerably more to the story, including and perhaps especially on the theoretical side.

This is really interesting! I ran into hints of this when I was doing the lit review for my dissertation. Not so much the internet part, but there was a RAND corporation memo on the Five-Year Plan of 1970 that mentioned the goal of placing all industrial production under rational, automatic control. I suppose I could have inferred an internet from that, but I was more interested in the idea that they wanted to run their economy with a giant linear program. (Which is an idea that was also briefly popular in the U.S. during the late 1940s and early 50s.)

Sevensor, I'd LOVE to read more about this. Is there more in your diss?

Not a whole lot. I mention it in passing. You can get the Rand corporation's 1970 Soviet Cybernetics Review here: http://www.rand.org/pubs/research_memoranda/RM6200z1.html It makes for fascinating reading.

Edit: If you're interested in the late-40s zeitgeist in the US, Tjalling Koopmans' 1951 Charnes Commission report is available here: http://cowles.yale.edu/sites/default/files/files/pub/mon/m13.... My apologies to Yale for hotlinking.

Edit 2: That's Cowles commission, not Charnes commission. Charnes might have been at the meeting (I forget off the top of my head) but it wasn't his commission!

Excellent! I know the 1970s RAND review and am delighted to learn about the Cowles commission material here.

In case you don't know it already, I heartily recommend "Red Plenty" by Francis Spufford.

It has also a dedicated website (http://www.redplenty.com/Red_Plenty/Front_page.html) so you can get an idea if this may be interesting for you.

Completely agreed. For anyone interested in the heyday of Soviet economics, please start with Spufford's fascinating and entertaining historical fiction! :)

Internet could not exist in a country whose existence was based on censorship and information control.

China seems to be handling it fine.

Head nodding Totalitarian regimes all around the world seem to be handling un-free networks just fine. So does the NSA. Networks do not necessarily embody our values. (Pardon the author rant!)

The Internet is a relatively new thing to China, so is GFW, the rapid upgrade of GFW, iirc, began appeared on somewhere after 2014. Before that, VPNs and Proxies were very stable, and minimal maintenance was required.

To the reason why it looks fine right now, my guess, is that the Chinese community is very much self-sufficient[1], plus, the majority of Chinese readers don't use any other language than Chinese itself. So it might not be about how it was handled, it's about what China is like.

-- [1] https://chinadigitaltimes.net/chinese/2016/04/%E5%A4%A7%E8%B...

Really interesting. "About what Russia is like" is the short story of this book, and there's huge room for more comparative work on the networks between these two giant socialist states without democracy on the Eurasian steppe. As best I can tell, this book may be translated into Mandarin before it makes it into Russian.

Yeah, and the ruling class in China is not capitalist at all. They are using driving forces of capitalism to achieve something else, and this is not buying Volvo production.

If by fine you mean blocking large chunks of information that could move the world forward all the while charging their tax payers for that privilege then sure it's being handled fine.

> Least of all did I imagine that this story would throw me headlong into a study of Soviet bureaucracies.

That is kind of like trying to understand the history of the Internet without understanding how the US Congress, DoD, (D)ARPA, and NSF relate and work.


From 6th paragraph:

> The first global civilian computer networks developed among cooperative capitalists, not among competitive socialists. The capitalists behaved like socialists while the socialists behaved like capitalists.

Great contrast! What is the deeper lesson here?

Thanks! There's many. Here's a couple: liberal economic language gets the cold war wrong. Our policies are often at their best when we start by doubting inherited commitments to either markets or states. Mixed economies have long been the default. And Hannah Arendt got something right: the cold war is not so much about a clash of competing ideologies as it is the rise of parties that privatize social (network) power. Etc.

Turns out GFW isn't so original after all. sigh

Ireland has been doing this for very long time

Ehm... The author of the article barely scratches the surface. He has no information on Soviet military networks, especially those decentralised ones introduced in anti-air missile system in the last decades of the USSR. КСА «Рапира-Ц» for those who really wants to know. And that guys, was the thingy. Also search for АСУ «Пирамида» if interested.

Unfortunately, that's how all discussions of Soviet technology inevitably end up: but ooh it's space or ooh it's military. This mindset is well captured in the famous track "Аквалангисты" by Mango-Mango.

Most of it was pretty mundane and touchingly backwards, just less shoddily built than civilian equipment. I took apart a fair quantity of BMP-2 and T-72 instrument panels; call it my contribution to arms reduction and the world peace. My first joystick was converted from an ATGM control station. I've been with ZGV in East Germany while it's equipment and supplies were hustled away by entrepreneurial NCOs and officers. I worked with some of the guys who copied ЕС ЭВМ series from IBM products with cracking stories about realities on the ground.

There's no secret sauce, despite no shortage of impressive Cyrillic acronyms and chest thumping.

So you admit, you never served you duty at ПВО. While you had been getting dirty selling diesel from T72 for Marlboro blocks in East Germany, other guys invented proper original communication systems and did their best. АСВТ designed and developed fully automatic Искра network, of which you probably never heard, listening to Mango Mango and other jazz.

"у нас есть такиииие приборы.. но мы вам о них не расскажем!"

Well, neither you served in air defence, so other than some bizarre personal attack am not sure what your point was here. It is however remarkable how you paint me a greedy traitor for bringing a dose of realism to your unrestrained pioneer enthusiasm.

I worked with a bunch of people from NII SA, the principal developer of ES series. Some of them ended up as mobile mounted C&C elements for the military and on relay com stations, including strategic networks. They were equally shoddy however (and clones of the Western architectures to begin with). The shortest guy in the team once had to sit inside a panel fanning an overheating assembly with a piece of plywood during military acceptance test.

Soivet computing in 1980s was pretty stale. And it simply doesn't happen that people making sad shit for civilian side become sudden geniuses when switching to defence orders. Computing systems were inadequate on factories, research bureaus, power grid, they simply could not not suck in the military too. There were no separate research institutes designing computer architectures for military only, no factories producing separate component base just for military. If anything, the civilian production was a side business at most enterprises and were manned by same people. General state of research, production culture and scientific effort determines the baseline, and USSR was behind on all three in CS.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact