For years, I remember Korean government had pushed ‘software engineer’ as a growing top job that can solve the insufficient labor demand for Korean GenZs.
The story ended up a spectacular failure for the following three reasons.
- LLM has been replacing junior devs
The first reason is common to all of us, across countries, but the other two may not be that common knowledge.
- Koreans cannot speak good English
- (Most) Dev skills are not unique, thus can easily be copied and replaced
A short tale from an OEM shoemaker for Nike
To help you understand what Dev people will experience, let me talk little bit about an OEM shoemaker story from Busan, Korea in 1999.
Back then, at the onset of China’s world debut as a ‘factory of the world’, Korean low-key labor markets were losing its advantage in noticeable speed. For a factory in Busan that makes Nike’s shoes, the company was confronted with Nike’s production base migration to China. One of the sales representatives (my relative) was ordered to go to South America (then it was Chile) for alternative sales lines.
The Chilean buyers did not care about the Busan factory’s production quality. They only wanted to buy Nike branded shoes at a bargain deal. A Chinese manufacturer made a deal with the Chilean buyers for selling Nike shoes on the condition that the sale would be kept secret from Nike. According to the Busan executive, when it comes to quality, the Chinese factory’s was somewhere in 1980s or early 1990s at best. He thought the Busan factory’s top quality shoes would be good enough to convince foreign buyers, but what the foreign buyers wanted was not the quality, but the brand.
The Busan factory left Korea and currently running the same operation in one of the ASEAN countries. (Can’t tell you more details for confidentiality.)
Skilled labor for shoemaker and for tech industry
Why the Busan factory lost its tie with Nike?
Because of Nike’s favor on cost over quality? Because of a Chinese firm’s malicious deal? Because of rogue business practice in South America?
Frankly, I don’t really think the quality was that different. For shoes, I do admit that there is a heavy portion of R&D needed, but for production, it’s all about efficiency maximization. I doubt the sales rep’s claim is really true. Nike is a world known brand, and if the quality changed, consumers must have raised concerns right away.
And business malpractice? Is it surprising? As far as I know, Korea also did the same practice in 1960s and 1970s. In fact, for such dark practices are common, and it was only a ticking bomb to blow.
Replaceable labor without branding
What was lacked to the Busan factory is branding.
The labor can easily be replaceable. In fact, in most manufacturing businesses, without branding, unless your company has the extreme quality control with incomparable production efficiency, competitors with cheaper labor cost will replace you sooner or later.
The same story is playing in the tech industry in these days.
Can dev work be branded?
Even OpenAI is struggling with fierce competition with Google’s Gemini and XAI’s Grok. There are open sourced based cheaper version like Chinese DeepSeek. Many of them are ready to de-throne OpenAI.
if you change the scope from company to AI scientists and AI engineers, unless one has irreplaceable skils, you don’t need to me to remind you that labor replacement can easily done at company level as much as you can one day switch from one LLM to another.
Except very few high quality AI scientists, almost everyone in that business is replaceable.
And, there are millions of devs who claim AI expertise but what they can do is just using OpenAI’s API through ready-made libraries provided by cloud services.
That skill can be learned within a few hours for some experienced devs. Because it’s just another coding library. Copy does not lose its value, and in fact, all library providers want many of you to use the library. They claim they are world-class dev, by the number of library adoption.
Journey to find cheaper labor
Knowing all this, would you hire a dev in Silicon Valley charing USD 300k, just because his previous employers are Google, Meta, and OpenAI? Or would you hire an Indian dev charing USD 3k for the same task, based on exactly the same dev library? Both of them may even have the common codebase.
Does the dev’s former employer carry any brading effect?
Inexperienced HR may give some credit for shiny and fancy names, but once you go to the production team, what’s needed is if you can deliver the product on time with the best possible quality.
And, the more power the production team has, the less likely the Silicon Valley dev will be able to find a job with requested salary.
After years of fooled choices, I can see many executives make right choices with better understanding of the software engineering business.
Then, why English matters?
Earlier, I mentioned
- Koreans cannot speak good English
for the reason that demand for decent jobs from Korean GenZs won’t be solved by training them as software engineers.
One may think that lack of English would disable them to access international labor market, while cheaper alternatives are already popular from India and most of 3rd world countries.
Yes, it is, but I advise you to check how those young Indian devs train themselves to be as good as Silicon Valley devs.
The lack of English vastly disables Korean junior devs accessing dev knowledge base that is changing in every hour. New tech stacks are offered constanty, and because of compatibilty, each one is forced to upgrade their libraries and systems to meet each other. Otherwise, your system becomes obsolete.
Two advanced economies experience exactly that situation.
Japan and Korea, lack of English kills the economy
In these two countries, dev people hardly read. In Japan/Korea, the high school education is divided into STEM/Non-STEM, and for STEM, only math solving skills are emphasized. Most dev’s background is from STEM, so they are trained to read short quetions and provide quick answers, instead of thinking logically for a long argument.
Poor reading and lack of English create very weird synergy. In most STEM heavy industries in Japan/Korea, it is forgivable to not read, especially if it is written in English.
I remember seeing my Korean dev team members running google translator on Stackoverflow posts. The translation is not always perfect, and they, for once, gave up fixing an error on the company website. They claimed that they could not find the cause, so they had to make the website from scratch, after two full day overnight try.
Making the same system again, which took 2 months?
I googled it in English, and found that it was due to Node.js’s one library’s minor update on a function’s name.
The team had spent two full days and nights. I fixed it in 5 mins.
Many Japanese and Korean devs claim that it is better not to upgrade anything. Because they know that the above case can happen.
There is a 10-year-old startup running a website made by Node.js version 6.x. I remember the version was released sometime in early 2016. Node.js version 6 reached its End-of-Life (EOL) in Apr 2019. They did ‘refactoring’ for a few times, but they must have been so afraid of learning new versions and its evolving ecosystem. Given that the EOL is 6 year ago, the company’s website has been exposed to outside attacks since then. i.e, it’s a ticking bomb.
In other countries, it is much easier to find a dev who not only is highly conversant with Node.js (or any JS based variants), but also is capable of migrating Node.js v6 programs to recent versions, like Node.js 2022 (LTS), if not, recent v24.
But, most documents are in English, and the competition within the dev market is sealed by language barrier, so, the devs are virtually living in Galapagos-like segregated eco-system.
While the countries had high domestic demand in IT, learning new is not needed that much. That’s what happened to old generation.
But, when companies are under extreme international competition and pressure to lower cost, while security concerns grow, it is inevitable that local buyers will no longer choose expensive yet old-fashioned domestic IT vendors.
Just like Nike had chosen a low-cost Chinese factory over Korean one, until Chinese factories lost its competitiveness after 10 years.
So, no branding, no job, if copiable.