South korean developer here. I am quite aware of government plans for software since I've done my service as a software developer for army.
There is a company called TmaxSoft which advertises itfself as only builders to the 'Korean OS'. Fueled by relatively high patriotism shared by average koreans, the government had been subsidizing this firm for more than 10 years. Despite the effort to liberate the nation from a private US firm this wasn't successful so far: first they tried to build their own kernel which froze at the public demo. After realizing it was dumb to write kernel from scratch, they began focusing on tuning the visual interface. The result was criticism that is was just a pathetic clone of MintOS(linux distro).
Due to bureaucratic debt over 10+ years, I think those government officials are desparately trying to meet an end to this mad project. The money and time they've spent well deserves a spot in governmental future plan in converting all american OS: Windows into Homemade korean government friendly OS.
This software firm builds many 'hard' softwares like OS, DBMS or even MapReduce data processing cluster(like hadoop and spark). One might naturally ask 'why build a wheel while there are plenty already built by open source developers?'
and economically aware person might also ask 'how are selling their products while competing against ubuntu, postgresql, spark?'
Cloning something from scratch and learning advanced technology in its process is a familiar thing in Korea. Enforcing home grown products to average customers, subsidizing one specific firm to take all domestic market share is how Samsung and Hyundai made its way to top global firms. So I guess the mental background for replacing OS by home made OS is similar to that which successfully grew Samsung. They are willing to pay off a lot of money so that a Korean firm outgrows Google, Facebook or Apple. An average developer knows it is a stupid plan to replicate linux or mysql from scratch. But government officials' heart still remain in the golden age when the very strategy they are executing with was valid for Samsung or Hyundai.
The way this firm hires its employees is tighly coupled with government. In our country all men are obliged to military service. However there are couple of options:
1. B.S holders: no exemption. 2 years in a random role.
2. M.S holders: you get to work for private firm for 3 years. Called 'special research agents'. Mostly work for engineering related firms.
3. Ph.D holders: you are exempted as soon as you become a Ph.D holder.
All these exemption rules according to the level of scholastic degree is to foster R&D capability of our nation. And here comes a perky question: 'What happens to Ph.D dropouts?' As some might have guessed: 'You go to TmaxSoft, get hefty salary, get a girl, get married.'
I wouldn't say much about it if there weren't a fact that really bugs me. They do hoard M.S holders from best colleges, however the graduate students they hire aren't limited to CS majors. I've seen multiple chemistry majors or industrial engineering majors who dropped out of Ph.D course doing software engineering there. I have no idea how they could 'train' a non CS major to write code for hard core stuffs like OS or DBMS. And they tell the government officials that they have the best pool of engineers from prestigious colleges, which would ease the government to making decision to subsidize this firm.
Er... I personally feel that this comment is too harsh to TmaxSoft and the government officials.
Yeah, TmaxSoft builds and sells compatible DBMS/parallel computing solutions to the government — but I feel that it’s entirely understandable to use domestic software for things like that. The killer features there wasn’t that it was domestic — it was that it’s cheaper that the alternatives and it’s compatible with the big ones, ensuring no vendor lock-in.
It’s definitely not a plan to make TmaxSoft some global company like Samsung or Hyundai; its not the 70s where the country has no big companies for the global market. The government has no incentives to do that, when Samsung’s software is used over the world. It’s just the preference of domestic software, and that TmaxSoft delivered actually usable software (not great, but is usable) to the government.
You know how crucial software gains reliability through large user community. I don't want to fall into some random bs while using their 'usable' DB. condolence to programmers involved to government projects which will enforce them to use 'usable' piece of framework from domestic firm.
If at some point a nationally important company wishes to pivot or at least threaten to pivot on some important tech, their statements will carry less weight unless they've already been developing the tech with some visible gusto. I think there is some strategic value to holding something like Bing, assuming its costs are reasonable.
This is not something unique to Korea most governments do this. Subsidise or promote a homegrown product so that the homegrown product becomes competitive in the future. NASA supporting private rocket firms is an example of this.
> Enforcing home grown products to average customers, subsidizing one specific firm to take all domestic market share is how Samsung and Hyundai made its way to top global firms.
I can see how this would ensure domestic success, but how does that translate to 'top global firm' status unless the products / features & pricing are significantly superior?
It basically gives domestic firms breathing room to survive and grow, versus being suffocated by the currently superior products or foreign firms. The hope then is that in the future, with a stronger financial footing, these firms' products will improve to be truly competitive with those of the global competitors.
At least in hardware/manufacturing, both South Korea and China have done this with great success thus far.
> It basically gives domestic firms breathing room to survive and grow, versus being suffocated by the currently superior products or foreign firms.
Yes, I get that. It's basic tariff protectionism. Many countries have done, and continue to, do this.
Here in Australia we've done it for decades, but with no global success.
> The hope then is that in the future ...
I hope that isn't the official strategy. ; |
The stronger financial footing (also achievable via vanilla domestic success sans tarrifs) doesn't necessarily guarantee global success, especially if other nation states are adopting similar fiscal & trade policies ... which it's fair to say they are.
> Examples? We have some of the most open markets in the world. We’re one of the only countries the US has a trade surplus with...
I guess the obvious example would be the car industry - mostly cars were just assembled in Australia in the most recent past, but previously there were some designed and built. Tariffs were applied to protect that industry (nominally the jobs associated with same) but despite the Australian car assembly sector basically being zero now, the tariffs still exist. It's almost like it's just a revenue generation scheme.
There's a long history of tariffs and duties within Australia [1] but none of those (to my knowledge) resulted in a thriving multi-national corporation (I'll exclude mining companies as they're primary industries, reliant on the lottery of resource availability rather than any particular commercial skill).
Perhaps Murdoch's News Limited, which undoubtedly obtained government support during its nascent era, but I'm not sure how comparable it is to the Samsung and Hyundais of the world (referring back to GP's claims).
Australia has it pretty hard be competitive exporting industrial goods since the supply chain is mostly overseas and the domestic market is small, which means most things will gehave to be exported again. All this results in high shipping costs and latency in addition to being a high wage country.
Back in the 70s and 80s there were many tariffs that were directly related to the manufacturing in Australia.
However, our tariffs were more about protecting jobs than protecting and developing industries. Australia has always been a "taker" when it comes to investment and we are shite at developing our own industries that aren't purely primary (ie extractive like mining or agricultural).
We have some companies today that are global but have nothing to do with tariffs. Atlassian in software, some mining related engineering and services companies.
There are others like CSL which was spun out of government, Brambles/CHEP that again, was spun out of government (CHEP stood for Commonwealth Handling Equipment Pool in WW2).
But our governments are terrible at promoting secondary industry or supporting them. Our R&D grants are complicated and are focused on compliance, not on results.
Australia is the "lucky country" but the full quote is:
"Australia is a lucky country run mainly by second rate people who share its luck. It lives on other people's ideas, and, although its ordinary people are adaptable, most of its leaders (in all fields) so lack curiosity about the events that surround them that they are often taken by surprise."
It sounds like Australia is a victim of the resource curse. Resources bring in the money regardless of how competitive you are. There are no incentives to compete and improve because Australia can afford complacency and incompetence.
Australia has a smaller population, very high wages and a technology-phobic government with policies that have filtered all investment into the property market for decades - rather than letting investment go into high-risk-high-reward projects. Not much to report on the megacorp front.
There are major Australian companies out there; the Atlassians, BHPs and CSL's of the world. They aren't Samsung by any stretch.
Australia has second rate government and second rate management. We're good at digging up dirt and shipping it, or ploughing dirt and shipping it in the form of food.
Our management and government are incredibly risk-averse and we follow trends and popular management theories instead of creating them.
No insight into Australian management, but having lots of valuable dirt to dig up does make it much harder to get going in other lines of business.
I think this is called "dutch disease" by economists: the high wages paid by the dirt-digging (early on) mean that you can't simultaneously be competitive in metal-bashing, which makes it hard to develop other industries.
A key idea here is that, to keep this support, the firms were also obliged to export. Even small numbers, at a loss, at first, but they had to have paying customers in America. Firms that couldn't do this were allowed to go bankrupt / get swallowed up.
At least in Korea. In less successful countries, domestic top dogs were simply protected, and were quite happy to keep making cars from the '50s.
An excellent book on this story is "How Asia Works" by Joe Studwell. (On which I'd be interested to hear any opinions from actual Koreans.)
That 'force overseas sales' thing sounds like a notable differentiator, though I'm not well versed enough to know if there's (m)any contra examples for this practice.
And thank you - I've added How Asia Works to my reading stack.
It's hard to get to the point where your products are significantly superior if you have no revenue while you experiment with improvements to your product until it can compete on its own. Funding doesn't guarantee success, but without it you may have no chance at all.
This was interesting to read. I had no idea that Korea was trying to move to their own platform, nor did I know it’s been in the works for a decade. Crazy!
> I have no idea how they could 'train' a non CS major to write code for hard core stuffs like OS or DBMS.
There is probably someone better fit to describe this [hah!] but schools aren't really designed to teach something specific. It is more about creating hard working drones. Doing anything meaningful in [say] chemistry is harder than brainfuck.
Korea doesn’t value software engineers like Silicon Valley does. Over there, it is considered menial work and the pay is VERY substandard. This has been true as recently as 2018, but it is (too slowly) starting to change. The only FAANG with an engineering office in Seoul is Google.
My terrible experiences with the remote-control app (https://play.google.com/store/apps/details?id=com.panasonic....) for my otherwise-fine mirrorless cameras now makes total sense. Even though the app has received constant updates for years, it's still extremely unreliable, and looks like hasn't had a visual update in ten years.
Language barriers, timezones (their night is our day; it is flipped). Most skilled Korean software engineers end up in the Valley. Remote development only goes so far.
I don't see it any differently than companies with an office in India. While English is not a national language in RoK, Koreans are still taught English early. In my limited experience, being married to a Korean and having worked with Indians, Korean professionals speak English about as well as Indian professionals.
The timezone is similar: India is 10 hours forward from America/New_York, RoK is 12 hours forward. Flights to RoK (12-14 hours) are actually significantly shorter than flights to India (20 hours).
Indians are probably paid less than Koreans, but I don't know. And I cannot speak to the technical proficiencies of either.
I'd definitely like to open an office in RoK some time.
If anyone on here has done so or is interested, send me an email.
Are you interested in moving to RoK?
Either way expect salaries at around 50-60k so not that cheap. And also the talent pool is rather small.
The English proficiency is almost non existent.
Hello, nice to meet another Korean (still wake at 4AM) on HN! :-)
> The result was criticism that is was just a pathetic clone of MintOS(linux distro).
I’m pretty sure you’re mixing up Harmonica with Tmax — Harmonica is based on Linux Mint (and is a pretty well operated open source project) while Tmax uses the Linux kernel, some BSD components and a custom desktop environment that features a rip-off interface of Windows.
There wasn't any english stuff for Tmax distro that I could find...
idk why they don't just adopt Debian + Gnome for their desktop stuff with some Korean development work on top of that. It's a great looking UI/UX and would be perfect especially if most of the software is via the browser/SaaS anyway. Or some occasional wine translations where needed.
But I guess that's just how governments work and their constant obsessions with the sunk cost fallacy.
Edit: Harmonica seems to be doing the right thing already, from the review:
> However, it is, at it's core, just Linux Minux with two special PPAs for Koreans on top.
This isn't something to shame them for, this is how it should be done.
> dk why they don't just adopt Debian + Gnome for their desktop stuff with some Korean development work on top of that. It's a great looking UI/UX and would be perfect especially if most of the software is via the browser/SaaS anyway. Or some occasional wine translations where needed.
Well, (as you mentioned in the edit) it doesn’t create something totally new — AFAIK it uses the cinnamon DE. Also, you mentioned about being ‘perfect’... but you will be surprised in the level of CJK support on any open source software. Even the big ones don’t support CJK well, unless the software uses the OS toolkit & operating system gives you for free — and Linux doesn’t. Software like Firefox’s CJK input is frequently broken in macOS (they don’t use Cocoa), and on most Linux distributions everything is just terrible (including Ubuntu). That’s the reason for using a local distribution.
> This isn't something to shame them for, this is how it should be done.
It could be great if we could just use Linux Mint or Ubuntu and call it a day; Linux people, please stop considering input managers and other UTF-8 stuff as bloat... or you will never get meaningful adoption to ordinary users in the CJK.
> Even the big ones don’t support CJK well, unless the software uses the OS toolkit & operating system gives you for free — and Linux doesn’t.
What specific CJK support is lacking on Linux? I only have experience with CJ on Ubuntu, where I didn't really find anything lacking. Noto fonts give sufficient Unicode coverage, everything speaks UTF-8 and fcitx is an adequate input method engine, even if its text prediction isn't Google-level smart.
There's no shame in trying, and no shame in pivoting after a cost/benefit analysis, and sticking with Windows.
In contrast, North Korea maintains its own Linux distro (https://en.wikipedia.org/wiki/Red_Star_OS#Version_4.0) and has for ten to twenty years, probably because they can't trust any proprietary software and they're not tied to Windows-happy U.S. DoD, and I would rather live in South Korea and use Windows than live in North Korea and use Linux.
> There's no shame in trying, and no shame in pivoting after a cost/benefit analysis, and sticking with Windows.
It wouldn’t been a shame if TmaxOS was advertised as a BSD distribution with Windows integration through Wine from the start — it advertises itself as a ‘pure korean OS’ without any mentions of BSD (which I’m pretty sure is a license violation) our Wine.
It later moved onto the Linux kernel and a more standard Linux environment (with some BSD components left over), and a custom DE, and is currently advertising itself as a Linux distribution. I’ve heard it is actually in a fairly useable state... but the OS already got a really bad rap.
Out of curiosity: why is it dumb to build a kernel from scratch? There are several such projects on Github with decent progress. Many of them are featured on HN time to time. Google also thinks that writing a new kernel is a great idea. I share their vision in this particular matter. We need a new kernel that was designed and implemented taking into consideration the last dew decades worth os research, especially into security and reliability.
It's pretty dumb for a consumer OS intended to run on commodity PCs, because the hard part is device drivers for all the hardware out there, and existing kernel projects are way ahead of you on that one. It might make sense in an environment where you have much more control over your hardware platform.
> first they tried to build their own kernel which froze at the public demo
Do you have a video of this? I found mention of it, and many videos of TmaxDay and the like; but without more context or speaking Korean, I'm finding it hard to locate.
One of the engineer was debugging the scheduler a day before demo day, poking here and there, inserting sleep(10); to random places. He merged his debug commit without deleting sleeps so the OS froze on demo.
what this implies: they don't have tests, no code review. And they're still developing OS.
I evaluated China's Red Flag Linux back in the day, around 2010.
They took a common linux distro, added Chinese fonts, and made a nice graphical login screen. I was impressed at how smooth the final result was actually.
If you're familiar with the final linux distros based on KDE (pre-Gnome 3), that's what it looked like.
So if you're a government IT staffer, study the history of Red Flag Linux.
Deepin a modern Chinese distro used in some gov i heard is also surprisingly good looking. Then there's WPS which now tends to ask me to login and whatnot a bit too much but i remember being delighted with it because back in the day it looked pretty and handled the garbage MS office stuff really well whereas libre and open office faltered here and there enough to be annoying.
>Despite the effort to liberate the nation from a private US firm this wasn't successful so far: first they tried to build their own kernel which froze at the public demo.
So? Linux, Windows, Mac OS, have all frozen during all kinds of public demos.
>After realizing it was dumb to write kernel from scratch
Is it? A nation state, especially one the size of Korea, should have no problem finding resources for writing a kernel from scratch - to a level suitable for running POSIX software and a modern UNIX-like userland.
To ensure foreign intelligence agencies have reduced visibility? To prop up local industry? Support customisations? Lots of valid reasons to roll your own OS, especially for National organisations.
There is a company called TmaxSoft which advertises itfself as only builders to the 'Korean OS'. Fueled by relatively high patriotism shared by average koreans, the government had been subsidizing this firm for more than 10 years. Despite the effort to liberate the nation from a private US firm this wasn't successful so far: first they tried to build their own kernel which froze at the public demo. After realizing it was dumb to write kernel from scratch, they began focusing on tuning the visual interface. The result was criticism that is was just a pathetic clone of MintOS(linux distro).
Due to bureaucratic debt over 10+ years, I think those government officials are desparately trying to meet an end to this mad project. The money and time they've spent well deserves a spot in governmental future plan in converting all american OS: Windows into Homemade korean government friendly OS.