Section3 Flashcards
1
Q
What is Profiting from Innovation (PFI) about?
A
- It’s about who profits from new inventions.
- Outlined by Teece (1986), focusing on:
- Appropriability regimes
- Dominant designs
- Complementary assets
2
Q
Define Appropriability Regime
A
- The extent to which an innovator can capture profits.
- Strong regime: hard to imitate (e.g., patents).
- Weak regime: easy to imitate (e.g., few legal protections).
3
Q
Why is Appropriability crucial?
What are the players?
A
- Determines who reaps most rewards: innovator, suppliers, or imitators.
- Impacts profit distribution among market players.
4
Q
What is a Dominant Design?
A
- The standard product or process format that emerges in an industry.
- Before it, competing designs (pre-paradigmatic); after it, industry converges.
5
Q
What are Complementary Assets?
and name the types
A
- Additional resources needed to commercialise an innovation effectively.
- Examples:
- Generic (e.g., cloud hosting)
- Specialised (custom APIs)
- Co-specialised (unique data sets).
6
Q
What if an innovator lacks Complementary Assets?
A
- May partner or contract with others (e.g., OpenAI + Microsoft).
- Risk of sharing profit or losing control.
7
Q
Is GPT3 easy to replicate?
A
- GPT2 was partially replicated from published details.
- GPT3 is harder to reverse-engineer due to secret training data.
- Shows a relatively strong appropriability (via secrecy).
8
Q
Has a dominant design emerged for large language models?
A
- Transformer architecture is now the core standard (e.g. GPT, BERT).
- Conversational interfaces (e.g. ChatGPT, Bing Copilot) are becoming mainstream.
9
Q
Why does OpenAI share profit with Microsoft?
A
- Lacks capital for its own data centres.
- Exclusive partnership secures computing power.
- Microsoft licenses OpenAI tech and co-develops infrastructure.
10
Q
According to Teece (1986), who profits from innovation?
A
It depends on:
- Appropriability regime (strong vs weak)
- Dominant design (pre vs post)
- Complementary assets (generic vs specific).
11
Q
Can weakening an appropriability regime ever be beneficial?
A
- Yes. Firms may deliberately open technology to prevent competitors’ patents.
- Example: Merck making gene data public to avoid gene patent blockages.
12
Q
Is Llama 2 fully open-source?
A
- Meta calls it open-source, but with restrictions (e.g., cannot improve rivals’ LLMs).
- Access often requires approval, so it’s not entirely public domain.