Well, because the way concepts show up are so context-dependent, it’s easier to compare by deep analogy to OTHER cases (that are concept instantiations) instead of by reasoning from underlying principle.
So Chang’s story is an opportunity to add to a store of cases in our heads.
Conversation
In Texas Instruments, Chang invented something called ‘learning curve pricing’ — that is, you price your new chips very low, in order to grab a large chunk of the market and drive volume. With that higher volume, you climb the learning curve and increase production yield.
1
3
Sure, you rack up a ton of losses at the start, but the low prices guarantee large market share, driving out competition. You also increase yields faster than your competition.
Later, when it becomes a trailing edge product, you recoup your up-front costs by continuing to sell.
1
3
Chang invented this in the 60s; TI became the world’s largest semiconductor on the back of the strategy. So how did they lose it?
The story is a little complicated, and you should listen to the full Acquired episode: acquired.fm/episodes/tsmc
To simplify …
1
1
The two broad reasons were:
1. TI tried to apply learning curve pricing to consumer products (LED watches + calculators).
2. IBM chose Intel for the IBM PC.
This is a simplification (there was also internal politics) so you should dig further + take that into account. But!
1
When consumer sentiment changed (people gave up on LED watches), TI was left with 100s of millions of dollars worth of inventory it had to write down.
And the growth of the PC industry drove volume to Intel’s x86 processors at a level way higher than TI’s production volume.
1
Ok, so what’s the lesson here?
Wrong question. What is the concept instantiation at play?
It is this: scale economies are great … if you can enjoy that scale over time. If consumer demand shifts away from your product, all the cost savings/yield you get from scale is useless.
1
4
Chang himself reflects on the fact that consumer products are different from pure semiconductor production, and a game he didn’t want to play.
And the second concept instantiation: learning economies kick in when you can ride a wave of production.
1
Sometimes a wave of production comes from slashing prices, taking market share away from your competitors in a rising market.
But other times the wave of production comes from adoption of a new game changing architecture. Like PCs and Intel’s x86 chips.
1
1
Chang started TSMC after he was put out to pasture at TI. What he did there is a reflection of his earlier lessons.
He positioned TSMC as a ‘pure-play foundry’ — and ONLY dealt in semiconductor designs and fabrication for other companies. No consumer products.
1
3
And he came back from retirement when he saw the mobile opportunity. TSMC used this wave of adoption to leapfrog Intel. Which has a neat analogy: it mirrored the x86 Intel-IBM PC story, where Intel leapfrogged TI thanks to the PC’s wave of adoption.
jamesallworth.medium.com/intels-disrupt
Replying to
I’m using this story to illustrate a point.
If you do ‘lessons learnt’, you’ll have to draw very narrow, context-specific lessons that might not transfer to new unique situations. We never step in the same river twice, etc.
Quote Tweet
Replying to @NeckarValue and @morganhousel
Asking the provocative question but since we never step in the same river twice esp for new areas like venture and increasingly equity investment - what is there to be gained by learning from history of another era? Esp if it turns out to limit your visions of new potentials
1
1
4
But if you ask: what concept is this an instantiation of, and what does this story highlight about the concept? Well, then you’ll build a set of cases that you can reference when thinking about a concept.
In this case it’s about what scale + learning economies looks like in use.
1
3
To recap:
Chang’s story is AN instantiation of scale economies.
‘Lessons learnt’ might force you to learn narrow lessons, over-fit to the specific moment in time.
‘Hunting for concept instantiations’ is a different way of reading history, that yields more timeless insight.
2
7
And the reason it yields more idempotent lessons is that it helps you reason better about how concepts ACTUALLY show up in the real world.
It grounds you in reality, so when you talk about ‘learning economies’ or ‘scale economies’ or wtv, you’re not talking out of your ass. /END
1
8