Patterns by kind of event

What confirmed winners had in common, split by kind of event — because a student hackathon, an open-source challenge and a NASA data challenge reward different things. Every share shows its sample size and a 95% range.

35confirmed winners and finalists

4kinds of event with winnersfrom 6 events with results, of 14 indexed

31%had no AI in the product95% range 19–48%

11%built custom hardware95% range 5–26%

What wins at each kind of event

Share of each kind's confirmed winners that showed a mechanism or played to a judging criterion. The thin line on each bar is its 95% range.

agency open data challenge (n = 20)
open source community (n = 8)
student major league (n = 6)

Show the numbers

Winning mechanisms
Winning mechanisms	agency open data challenge (n = 20)	open source community (n = 8)	student major league (n = 6)	sponsor platform (n = 1)
visible engineering depth	50% 10/20 · 30–70%	50% 4/8 · 22–78%	67% 4/6 · 30–90%	0% 0/1 · 0–79%
complete submission package	35% 7/20 · 18–57%	88% 7/8 · 53–98%	0% 0/6 · 0–39%	0% 0/1 · 0–79%
expert data for everyone	60% 12/20 · 39–78%	0% 0/8 · 0–32%	0% 0/6 · 0–39%	0% 0/1 · 0–79%
human stakes	30% 6/20 · 15–52%	13% 1/8 · 2–47%	33% 2/6 · 10–70%	0% 0/1 · 0–79%
one sentence clarity	25% 5/20 · 11–47%	0% 0/8 · 0–32%	67% 4/6 · 30–90%	0% 0/1 · 0–79%
trend wave timing	25% 5/20 · 11–47%	38% 3/8 · 14–69%	0% 0/6 · 0–39%	0% 0/1 · 0–79%
sponsor platform mastery	0% 0/20 · 0–16%	63% 5/8 · 31–86%	33% 2/6 · 10–70%	100% 1/1 · 21–100%
demo first wow	0% 0/20 · 0–16%	13% 1/8 · 2–47%	67% 4/6 · 30–90%	0% 0/1 · 0–79%
cultural reference hook	0% 0/20 · 0–16%	0% 0/8 · 0–32%	67% 4/6 · 30–90%	0% 0/1 · 0–79%
physical artifact	0% 0/20 · 0–16%	0% 0/8 · 0–32%	67% 4/6 · 30–90%	0% 0/1 · 0–79%

sponsor platform (n = 1) — too few winners to chart; shown here only.

Show the numbers

Judging criteria
Judging criteria	agency open data challenge (n = 20)	open source community (n = 8)	student major league (n = 6)	sponsor platform (n = 1)
technical depth	40% 8/20 · 22–61%	63% 5/8 · 31–86%	67% 4/6 · 30–90%	0% 0/1 · 0–79%
impact	45% 9/20 · 26–66%	63% 5/8 · 31–86%	33% 2/6 · 10–70%	100% 1/1 · 21–100%
design ux	55% 11/20 · 34–74%	13% 1/8 · 2–47%	0% 0/6 · 0–39%	0% 0/1 · 0–79%
challenge relevance	55% 11/20 · 34–74%	0% 0/8 · 0–32%	0% 0/6 · 0–39%	0% 0/1 · 0–79%
scientific validity	40% 8/20 · 22–61%	0% 0/8 · 0–32%	0% 0/6 · 0–39%	0% 0/1 · 0–79%
originality	20% 4/20 · 8–42%	0% 0/8 · 0–32%	50% 3/6 · 19–81%	0% 0/1 · 0–79%
wow factor	0% 0/20 · 0–16%	13% 1/8 · 2–47%	67% 4/6 · 30–90%	0% 0/1 · 0–79%
sponsor api use	0% 0/20 · 0–16%	38% 3/8 · 14–69%	17% 1/6 · 3–56%	100% 1/1 · 21–100%

AI and hardware among winners

Does winning take AI or a physical build? It depends on the event.

No AI in the product

Custom hardware

Show the numbers

Kind of event	No AI in the product	Custom hardware
agency open data challenge (n = 20)	45% 9/20 · 26–66%	0% 0/20 · 0–16%
open source community (n = 8)	0% 0/8 · 0–32%	0% 0/8 · 0–32%
student major league (n = 6)	33% 2/6 · 10–70%	67% 4/6 · 30–90%
sponsor platform (n = 1)	0% 0/1 · 0–79%	0% 0/1 · 0–79%

How much to trust these numbers

Every share has a 95% range, because the samples are small. Example: 4 of 6 student winners brought a physical artifact: 67%, but the real share could plausibly be anywhere from 30% to 90%.
Kinds of event with fewer than five winners are shown in the tables but not charted.
Mechanisms and judging criteria are mostly the curator's reading of each project; every record says who made each claim.
Most events are indexed with winners only, so these charts show what winners did — not how often an idea wins. Base rates by event ↓
We tested this method on NASA Space Apps 2024 winners it had never seen: the event-kind playbook partly held, and the recommended idea shape matched 2 of 10 winners. Read the backtest ↗

No confirmed winners yet

startup accelerator — Platanus Hack 24 (inferred), Platanus Hack 25 (ft. Buk), Platanus Hack 26 Bogotá (LATAM Tour), Platanus Hack 26 Buenos Aires (LATAM Tour)
invite residential — SOON past edition (undated)

Saturation by event

Base rates exist only where every submission is indexed.

Event	Indexed	Roster	Top domains
Platanus Hack 26 Buenos Aires (LATAM Tour)	25 / 25	Full · base rate	ai security 6 · agentic payments 5 · identity provenance 4 · agent infrastructure 3 · dev tools 3
Platanus Hack 26 Bogotá (LATAM Tour)	24 / 24	Full · base rate	emergencies disaster 7 · simulation forecasting 7 · ai security 3 · health 3 · legal access 3
Croma GOV-TECH AI Hackathon 2026	12 / 12	Full · base rate	due diligence risk 6 · govtech transparency 6 · commerce retail 2 · fintech 1 · legal access 1
Platanus Hack 25 (ft. Buk)	14	partial	education 4 · commerce retail 3 · security networking 2 · social connection 2 · accessibility 1
NASA International Space Apps Challenge 2024	10	winners only	earth observation 5 · agriculture food 3 · climate sustainability 3 · space science 3 · creative media 2
NASA International Space Apps Challenge 2025	10 / 18,860	winners only	space operations 4 · space science 3 · climate sustainability 2 · earth observation 2 · education 2
Platanus Hack 24 (inferred)	10	partial	education 3 · fintech 3 · dev tools 2 · memory journaling 2 · security networking 2
Hackathon for MCP & AI Agents 2026	8	winners only	agent infrastructure 4 · ai security 3 · cloud native infra 3 · accessibility 1 · commerce retail 1
SOON past edition (undated)	5 / 17	partial	marketing growth 3 · creative media 2 · games entertainment 2 · dev tools 1 · memory journaling 1
Hack the North 2025	4	partial	robotics hardware 3 · accessibility 1 · careers hr 1 · creative media 1 · education 1
UofTHacks 11	2	partial	dev tools 1 · productivity 1 · robotics hardware 1 · social connection 1