technologybriefs
10:14in productionCh. 1 · What it is/ 10:14 · ceiling 15 min
Models

Relational model

Codd · 1969

Codd’s relational model is not a database product—it’s the only formal specification that makes 'consistent data' a logical, not just operational, claim.

Codd’s 1969 relational model is a formal specification—not software, not SQL, not a product. It defines data as tuples in relations, enforces consistency via logic, and insists all information be explicit. Real databases violate it daily. That gap is where data engineering begins.

ModelsAbstractionAlgorithmsComputingAbstraction across the network →
Chapters & takeaways4
  1. 0:52
    What it is

    It is a logic-based general model of data—not an implementation, but a formal specification rooted in predicate logic.

  2. 2:28
    The Information Principle

    All data lives in tuples; nothing else counts. Consistency is decided by evaluating boolean constraints over those tuples.

  3. 4:08
    Keys, precisely defined

    A candidate key is the smallest unique attribute set. A foreign key is a projection subset—not a pointer, not a string match.

  4. 5:58
    Missing information, formally

    Codd extended it twice to handle missing data—not with hacks, but with multi-valued logics: three-valued, then four-valued.

Worth your time?

Yes. Study the whole thing.

4.5/ 5
What works
  • technology/models
  • technology/software
  • technology/systems
  • technology/tech-history
What does not
  • technology/failures
Study it if
  • data-engineers
  • database-designers
  • logic-programmers
Skip it if
  • application-developers-who-only-use-orms
  • no-code-tool-users
The written brief1 min read

What it is and the problem it solves

It is a formal, logic-based data management framework. It solves the problem of unstructured, implementation-dependent data access by grounding databases in set theory, relational algebra, and predicate logic.

How it works

It represents all data as tuples in relations, governed by first-order predicate logic. It enforces integrity through candidate keys, foreign keys defined by subset-of-projection correspondence, and arbitrary boolean constraints. It adheres strictly to the Information Principle: no data exists outside explicit tuple values.

What works

Its constraint model works: boolean expressions can fully define consistency. Its key definitions work: candidate keys uniquely identify tuples; foreign keys enforce referential integrity via projection subsets. Its Information Principle works as a design invariant.

What does not

Real-world SQL systems violate its core tenets: they permit duplicate rows, rely on positional ordering, and treat NULL as a marker rather than a logical value—despite Codd’s later three- and four-valued logics for missing information.

What it changes

It replaced ad hoc navigational access with declarative, logic-based querying. It shifted database design from physical storage layouts to semantic modelling of relationships—and made data consistency a matter of logical entailment, not programmer discipline.

Is it worth your time

Yes—if you design, validate, or critique data systems. Its logic remains the only formal benchmark against which real-world SQL databases are measured for correctness, consistency, and expressiveness.

Same field · Models4 of 6
9:18
Diffusion modelDiffusion models are a 2015 method for sampling from complex distributions using thermodynamic-inspired noise reversal. They work by learning to undo a diffusion process—turning noise into data, step by step. They succeed in computer vision tasks as of 2024. They do not generalise beyond vision. They require iterative sampling, trading speed for quality. Their mechanism is physical in inspiration, statistical in execution, and narrow in current utility.
8:14
Gaussian splattingGaussian splatting is a volume rendering technique introduced by Lee Westover in the early 1990s and revitalised in 2023 by Inria for real-time radiance field rendering. It uses differentiable, anisotropic, explicit 3D Gaussians integrated from sparse camera-calibrated points. It is rasterisation-based, GPU-optimised, and avoids surface or line primitives. It enables fast photorealistic reconstruction from few images—but offers no mesh output, no topological guarantees, and limited editability.
9:11
Generative adversarial network2014GANs are an adversarial framework for unsupervised generation. They work by pitting two neural networks against each other: a generator learns to produce data that fools a discriminator, which itself improves at spotting fakes. This indirect training yields samples statistically aligned with the training set—including photorealistic images that pass superficial human inspection. But GANs offer no guarantees on semantic coherence, stability, or interpretability. Their value lies in enabling distribution learning without labels—not in reliability, control, or fidelity.
9:28
LoRA (machine learning)LoRA makes fine-tuning large language models radically cheaper to train without slowing inference—by replacing full-weight updates with low-rank matrix additions. It works. It scales. It merges. But it assumes weight updates are low-rank—and that assumption fails where tasks demand dense, global parameter shifts.
Up next in Technology

Cobot

· 9:49

‘Cobot’ isn’t a robot type—it’s a risk-assessment label you have to verify yourself.

Play nextBack to Models
9:49