4 Comments
User's avatar
Michael Schrage's avatar

just want to say this an excellent piece that i now discover preceded my own...my one pushback: 'trustworthiness' in vibe contexts likely overrated....much as jazz and improvisational 'second-city'-like humor can prove discordant and 'unfunny,' vibed analytic 'insights' can miss the mark while still being usefully revealing....the larger/more important thesis: talented 'prompters/prodders/interrogators/questioners' with a sense of whimsy, curiosity and/or adventurousness are sure to [sorry] 'generate' all manner of effective/affective 'insights' vibing with typical/atypical dataset combinations....my bet: 'vibe analytics' will rival 'vibe coding' in value creation and global impact by 2026's end....indeed, 'vibe analytics' and 'vibe coding' will blur and blend in remarkable ways....

Expand full comment
Yoni Leitersdorf's avatar

Thank you for the kind words :) I do agree, vibe analytics may have some value through prodding, interrogation, etc...

But in that case, isn't it just EDA?

Expand full comment
Michael Schrage's avatar

ha! sadly, neither fred mosteller nor john tukey are around to answer that question...so i took the liberty of asking chatgpt on their behalf:

**Vibe analytics** and **exploratory data analysis (EDA)** both prioritize discovery over confirmation—but they diverge in *epistemology, interaction, and intent*.

* **EDA**, as Tukey envisioned, is a human-led, visual-statistical interrogation of data to surface structure, spot anomalies, and guide formal modeling. It’s disciplined improvisation, grounded in tools.

* **Vibe analytics**, by contrast, is **machine-assisted semantic improvisation**—where prompts, context windows, and large models collaborate with the analyst to iteratively **reframe what matters**, not just explore what’s there. It doesn’t just ask *what the data says*—it asks *what the analyst is trying to mean*, and adapts accordingly.

> EDA finds structure in data.

> **Vibe analytics co-evolves structure and salience—until insight feels intelligible.**

Tukey would respect the iterative, question-first ethos.

**Mosteller** would admire the system's ability to **surface tacit assumptions and latent priorities** in real time.

-

Expand full comment
Michael Schrage's avatar

linked in won't let us connect so may i ask you to send (@ my mit address), an address where i can sends/share my mit sloan 'vibe analytics' piece....?

my thanks - and apologies! - i advance... cheers...

Expand full comment