Notebooks 2026 — Jupyter, Marimo, Observable

[khalilbenaz]

---

lede: Jupyter règne mais Marimo (reactive, OSS) et Observable challengent. Patterns pour data science et exploration en 2026.
lede_en: Jupyter rules but Marimo (reactive, OSS) and Observable challenge. Patterns for data science and exploration in 2026.
title_en: Notebooks 2026 — Jupyter, Marimo, Observable

Le paysage notebooks 2026

• Jupyter : leader historique, écosystème massif.
• Marimo : OSS, reactive (à la Excel), Python.
• Observable : JS, reactive, sweet spot dataviz.
• Quarto : reproducible research, multi-langue.

Jupyter — la référence

Jupyter Notebooks restent le défaut data science. Écosystème :

• JupyterLab interface.
• ipywidgets pour UI.
• Voilà pour publier comme app.
• nbconvert pour exports.

Limites : ordre d'exécution mutable cause bugs ("hidden state"). Mauvaise reproducibility.

Marimo — le challenger reactive

Marimo (lancé 2024) répond aux limites Jupyter :

import marimo as mo

@mo.cell
def cell1():
    x = 10
    return x,

@mo.cell
def cell2(x):
    y = x * 2
    return y,

Reactivity : modifier x → cell2 re-exécute auto.

Avantages :

• Pas de hidden state : pure data flow.
• Reproductible : ordre d'exécution déterministe.
• Stored as pur Python : git-friendly, pas JSON.
• Apps : déployer notebook comme app interactive.

Observable — pour dataviz JS

Observable (web-based) sweet spot :

• D3 / Plot integration native.
• Reactive cells JS.
• Partage facile via URL.
• Idéal explanatory dataviz.

Quarto — reproducible research

Quarto unifie Markdown + code (Python/R/Julia/Observable) :

---
title: My Report
format: html
---

## Analysis

\`\`\`{python}
import pandas as pd
df = pd.read_csv("data.csv")
df.head()
\`\`\`

Render to HTML, PDF, slides. Pour rapports techniques formels.

Choix selon usage

	Sweet spot
Jupyter	Exploration libre, ML training
Marimo	Apps interactives Python, prod-ish
Observable	Dataviz JS publiable web
Quarto	Rapports, papers, docs technique

Verdict

Jupyter reste défaut pour exploration. Marimo gagne du terrain pour apps Python interactives. Observable pour dataviz web. Quarto pour reports formels.

Pour 2026, ouvrir le toolkit au-delà de Jupyter seul.

2026 notebooks landscape

• Jupyter: historic leader, massive ecosystem.
• Marimo: OSS, reactive (Excel-like), Python.
• Observable: JS, reactive, dataviz sweet spot.
• Quarto: reproducible research, multi-language.

Jupyter — the reference

Jupyter Notebooks remain data science default. Ecosystem:

• JupyterLab interface.
• ipywidgets for UI.
• Voilà to publish as app.
• nbconvert for exports.

Limits: mutable execution order causes bugs ("hidden state"). Poor reproducibility.

Marimo — the reactive challenger

Marimo (launched 2024) addresses Jupyter limits:

import marimo as mo

@mo.cell
def cell1():
    x = 10
    return x,

@mo.cell
def cell2(x):
    y = x * 2
    return y,

Reactivity: modify x → cell2 auto re-executes.

Advantages:

• No hidden state: pure data flow.
• Reproducible: deterministic execution order.
• Stored as pure Python: git-friendly, not JSON.
• Apps: deploy notebook as interactive app.

Observable — for JS dataviz

Observable (web-based) sweet spot:

• Native D3 / Plot integration.
• JS reactive cells.
• Easy sharing via URL.
• Ideal for explanatory dataviz.

Quarto — reproducible research

Quarto unifies Markdown + code (Python/R/Julia/Observable):

---
title: My Report
format: html
---

## Analysis

\`\`\`{python}
import pandas as pd
df = pd.read_csv("data.csv")
df.head()
\`\`\`

Render to HTML, PDF, slides. For formal technical reports.

Choice by use

	Sweet spot
Jupyter	Free exploration, ML training
Marimo	Interactive Python apps, prod-ish
Observable	Web-publishable JS dataviz
Quarto	Reports, papers, technical docs

Verdict

Jupyter remains default for exploration. Marimo gains ground for interactive Python apps. Observable for web dataviz. Quarto for formal reports.

For 2026, open toolkit beyond Jupyter alone.

[MikeGarciaAGM]

Implemented and opened PR for this issue: #348

Scope:

• add issue content to posts.json as a new post entry
• preserve existing structure/order