I have been coding for 24 years, and the skills I relied on for a living have been completely wiped out four times.
Shen Translation is the translation team under 36Kr, focusing on technology, business, workplace, life and other fields, focusing on introducing foreign new technologies, new perspectives and new trends.
Editor's Note: In the past, technological reshuffles only deprived people of tools, while AI directly targets the intellectual core that human beings take pride in. Notes from a 24-year veteran in the industry: Hard skills will eventually be reset to zero, and only the failures you have personally gone through are the intuition that machines cannot steal. This article comes from translation.
The Morning When I Felt My Work Was Worthless
When I woke up this morning, I poured a cup of coffee and casually browsed a summary of the latest AI research progress.
That summary was not written by me, but completed by an AI agent. While I was fast asleep, it ran automatically, read more papers than I could read in a week, and organized the truly core content into a neat and clear list.
I held the coffee cup and sat there in a daze for a moment.
Twenty-four years ago, people paid me because I could memorize knowledge; but today, people pay me just to read the things that the machine has memorized for me.
This emotion has its own exclusive name, and I believe many people have felt it deeply recently. It is between relief and desolation — your efficiency is higher than ever before, but your sense of inner insignificance is also stronger than ever, and the two happen to be true at the same time.
But one thing kept me calm.
This is already the fourth time.
In 2002, I started writing mobile phone software at Motorola. Since then, I have witnessed my profession being sentenced to "death" four times. In the first three times, those who issued the death notice saw the drastic changes in technology, but underestimated the resilience of people. I'm not sure if the fourth time will be different. But I know deeply what the first three big reshuffles taught me — no one has ever written these truths into blog posts, so I plan to write them out.
Every big reshuffle I have experienced, at the moment when I was in it, looked like a complete doomsday.
The First Reshuffle: Back Then The Rule Was That Everything Had To Be Stored In Your Mind (2002)
When I first started in the industry, the industry rules were simple and clear, and everyone knew them:
You have to keep knowledge in your mind, not rely on the tabs in your browser.
Looking up other people's code on the Internet would be regarded as a defect in ability or even character. If someone caught you searching for answers online, their reaction would never be "You are so fast, you found it so quickly", but "How could you not even know this?" The standard to measure the value of an engineer completely depends on what you can write from memory in front of a whiteboard without any external help.
At that time, Stack Overflow had not yet been born — it was not launched until September 2008, six years later. Even if you wanted to take a shortcut, there was nowhere to go.
So we had to memorize them by force. We repeatedly studied the technical manuals of the chips until we could visualize the specific details in our minds with our eyes closed; there were printed reference manuals on the desk, covered with coffee stain rings. Senior engineers were respected entirely because their brains could hold a huge system architecture at the same time.
However, as the Internet was flooded with all kinds of ready-made answers, the long-standing value evaluation system collapsed overnight.
No one would send a special email to announce the end of this era. The year before, searching online was still an embarrassing thing; just a few years later, engineers who refused to search would only be regarded as procrastinators.
I witnessed many excellent colleagues find it hard to accept this gap. They spent a whole decade building their professional reputation with amazing memory, but the market no longer paid for this skill in the twinkling of an eye. Some of them chose to adapt; others secretly resented, and even accused young engineers of not being real programmers.
During this year, I often thought of those grumpy seniors. Because today, I can finally empathize. They were keenly aware that something precious was passing away, and that was not wrong; what was wrong was that they took the wrong path after facing the huge change.
The first skill I made a living from in my career was memory. When the market no longer paid for it, I didn't even notice that its funeral had been held quietly.
The Second Reshuffle: A Mobile Phone That Ended Our Feast (2007)
In January 2007, Apple launched the first generation iPhone.
I want everyone to feel what the world looked like inside a top mobile phone manufacturer before that. At that time, Motorola RAZR (the Razr series) had sold a total of about 130 million units, making it the best-selling flip phone in history. When you stand behind such a huge successful product, the company's building seems indestructible, and the corridors seem to last forever. The whole logic of the software you develop seems to be the only correct way in the world — after all, the brilliant data is there.
However, the advent of just one product completely shook the entire foundation.
The ending is already known to everyone. On January 4, 2011, Motorola was split into two parts; in May 2012, Google acquired Motorola Mobility for 12.5 billion US dollars, and then sold it to Lenovo for about 2.9 billion US dollars in October 2014.
I went through the entire transition phase completely, which is exactly the part that the outside world rarely mentions. The process was not thrilling, and it was full of endless meetings; it was the compromise that all the participants kept silent even though they knew the product roadmap was full of loopholes, because speaking up would cost their jobs; it was watching a group of top smart people trying their best to defend the decisions they no longer believed in half a year ago.
It was there that I realized the principle that I still believe in to this day:
Change is terrifying, change inevitably breeds failure, and failure is your most precious data.
This is not the kind of glamorous data that can be written in a report PPT, but costly real money — it may cost a supervisor a quarter's performance, and another key member the entire team. This kind of data reveals the underlying truth of how something operates in reality; it cannot be bought with money, nor can it be derived from algorithm models. You can only truly learn it by being in the vortex of defeat and observing intently when it happens.
Witnessing the collapse of an industry, I completed the most unforgettable further study. It took me several years, but every minute and every second is priceless, because now I can catch the signs of the building collapsing in advance. Today, as soon as I walk into a meeting room and feel that strange silence, I know it clearly — because I have experienced this feeling long ago.
Failure is extremely expensive data, and your career is your receipt.
The Third Reshuffle: A Quiet Transformation That No One Announced (2010s)
This third change didn't even hold any press conference.
Mobile terminals were no longer the forefront of innovation. Massive data and machine learning gradually took over the stage and became the soil for nurturing core problems. There were no keynote speeches, no epoch-making products, no clear dividing line between the old and the new, and everything became reality quietly.
In this baptism, I keenly observed a behavior pattern that has been repeated in every subsequent change:
People who see the general trend are full of panic, and because of the panic, they choose to act immediately. They use their nights to catch up on new knowledge without support, learning with faltering steps, accompanied by huge frustration and self-doubt throughout the process. It is fear that drives them to move forward.
People who can't understand the trend are carefree. They have no interest, point out the country without curiosity, and just stay in place and watch quietly.
When the dust settles and the trend is beyond doubt, it is exactly these people who jump out and rant that the whole thing is a mistake from beginning to end. This is by no means a rational judgment derived from in-depth review, but simply because they have fallen behind — after all, dismissing the general trend as a "mistake" is much more decent than admitting that you lack foresight.
This scene is still happening repeatedly. Right now, around you, all the voices around AI are like this. I have seen it many times in the comment area, meeting rooms and internal group chats of the team.
Putting aside the embellishment, this observation has a conclusion that makes people feel stuck in the throat, even applicable to myself:
Feeling fear just means you have seen the situation clearly. Facing the current AI wave, if you don't have any waves, it's not a sign of strong inner peace. The apparent calm may mean that you have a careful plan long ago, but more likely it is because you didn't open your eyes to see the world at all.
People with awe are already running, while carefree people are waiting and watching; when they are completely left behind by the times, they will attribute all this to a ridiculous farce.
The Fourth Reshuffle: Waking Up And Finding You Have Been Left Behind
This brings me back to the cup of coffee in the morning.
Now I am developing AI and machine learning software. On most days, I still go deep into the game and write code personally. If I give a real report from the front line, it is:
You have to learn every single day. You can't stop for a day. In this field, there is no fluke of "learn once and lie flat for three years".
This is not an inspiring quote to boost morale, but a complete complaint, and that's exactly what I think. This state is too torturous. Now my mornings start with the summary of night dynamics sorted out by the agent, because only in this way can I not be left behind by the times. The speed of technological iteration has long exceeded the reading limit of a single human being, so that I have to use machines to catch up with the pace of machine development.
On some mornings, this makes me feel like I have superpowers.
But on other mornings, it's like running nonstop on a treadmill whose speed is completely controlled by others.
Both feelings are extremely real, and I have long stopped trying to choose one or the other. I think that forcing people to have only one emotion towards AI is the root cause that makes everyone physically and mentally exhausted. You can completely open the results generated by the agent, while marveling "This is amazing", and secretly thinking in your heart "I really miss the days when I did all this by myself".
The core that makes this big reshuffle different from the previous three does not lie in the speed of iteration, but in the target it impacts.
The first three subversions deprived me of my tools; this time, it directly aimed at the core part that I once thought represented my self-worth.
Relying on machines every morning to catch up with the evolution of machines — this is the whole truth of work now.
This Melancholy Is Real, You Can Feel It Fully
Recently I read an article that expressed this feeling more accurately than I did.
Programmer Andy Balaam published an article, candidly talking about the loneliness and sadness that AI brought to him. What touched me was not his worry about unemployment, but the clear notification — the things you love, the skills you have spent half your life polishing to perfection, are now regarded as cheap standard products everywhere. He pointed out that this craft used to be his spiritual totem, the foundation for him to prove his self-worth; watching the whole industry shrug it off and ignore it is itself an indescribable trauma. At the same time, he also put forward a view that surprised me: many professions in history have long suffered this kind of objectification treatment for many years — farmers, nurses, drivers, teachers, cleaners, caregivers. Some of us are horrified to face being replaced, largely just horrified that we are unexpectedly included in this list.
These words are weighty and hit the nail on the head.
The subsequent discussion on Hacker News is also thought-provoking, the main reason is that everyone is expressing different opinions frankly. A netizen left a message saying that what really makes people feel pain is not the current situation, but the development direction of technology running wild — because if progress stops at this second, the panic in people's hearts will dissipate. Another netizen confessed that he used to pin his self-worth on writing exquisite and clever code, and now this impact on his cognition is embarrassing, but perhaps it is time to break the situation long ago. Others described a normal work state that I hate the most: daily work is completely reduced to layer upon layer of review of the solutions and products generated by the machine, while the part of deep thinking that people really love and indulge in is completely handed over.
I have experienced those days. On that day, I did far more reviews than actual construction. It tasted really tasteless.
Therefore, I will never persuade you to force a smile and cheer up. Three vicissitudes of technology have not taught me to skip sadness; on the contrary, they tell me that deliberately suppressing and avoiding this emotion will exactly make you become those grumpy senior engineers in 2006 — pointing fingers at the younger generations all day long, chattering and scolding that they are not real programmers at all.
You have to face and fully experience this loss, and then you are qualified to decide where to go next.
The pain you deliberately avoid will not dissipate out of thin air, it will only quietly transform into arrogance and contempt for later comers.
After Four Big Disasters, Why Did I Survive?
What I'm going to say next may be counter-intuitive, and I went through four big waves before I finally realized it:
None of my hard technical strengths survived, none of them made it through.
The rote memorization that I relied on in 2002 has long gone; embedded mobile phone development has also become a historical relic; half of the skills learned in the big data era have long been replaced by more powerful modern tools. If I measure my career purely by a specific hard skill I have mastered, I have been completely reset to zero four times, and at this moment I am still empty-handed, back to the starting point.
However, there is indeed something that has survived across the times, and it becomes more and more valuable every time.
Two months before the project schedule showed flaws, I could keenly smell the signs of project failure; I knew which meeting was the core game that really settled the overall situation; I understood the gap between a demo prototype that can run and a truly stable industrial-grade system — and this cognition was tempered in the cruel actual combat where I delivered the software to millions of physical mobile phones, and you were not allowed to play a hotfix casually on Tuesday.
I learned how to swallow the bitter fruit of making mistakes in full view of the public and still move forward firmly. This is the lesson taught to me by the second big reshuffle, and it is also the lesson that cost me the most.
I know how to start all over again in the embarrassment of feeling stupid, because I have repeated such mistakes three times; and now, I have realized that feeling stupid is just the first step in deep learning, not a proof that you have come to the end of the road.
These are not specific codes, but all of them are precipitated from the actual combat of writing code personally.
The most wonderful thing is this: without practical operation, you can never get this precipitated aesthetic taste and insight for no reason. There is no shortcut in the world that allows you to skip 20 years of underlying construction and directly jump to the state of having sophisticated intuition. Now AI agents can write code much faster than me, but at the time when the building was collapsing in 2008, the agents were not sitting in the meeting room where the seemingly indestructible business empire was falling apart.
This is the value I really rely on now. It took me four complete destructions and rebirths to suddenly realize that I have actually been selling this precipitation all the time.
Every specific technology I have mastered has become history, but I have firmly kept everything that those technologies once gave me.
The Unknown That Still Makes Me Awe-struck
I once told myself that I must not perfunctorily end with a set of happy words, so the following is the dilemma that I can't solve yet:
The first three technological reshuffles innovated tools, and human beings were still the masters in the decision-making chain; this time, it directly aimed at the whole chain itself.
If technology continues to evolve at the current wild speed, I can't assert how much value the human pride in intuition and decision-making can retain, or how long it can last. Anyone who swears to know the future is essentially just groping for an elephant in the dark. The confident optimists are making arbitrary judgments out of thin air, and the alarmist doomsday theorists are also speculating. At this moment, the two sides who shout the loudest are the voices that are the least worthy of listening to.
What I can give is never