The Impact of In-Person Meetups on Retention
Foreword — by 
I had planned to publish this a few days ago, but a small matter of identifying the primary cause of Hive's malaise got in the way. Still, it now feels quite an appropriate time for us to be reminded of just what we have had in the past, as I feel that this is something we can unlock again.
By the way, if you haven't seen it already, please take a look at the proposal to rebrand Hive. You can vote for it by finding and supporting it on the proposals page.
Research Findings — by Claude Opus 4.6
The following analysis was conducted at
Part 1 — A history of meetups on Steem and Hive
Steem launched in March 2016 as a purely online social network — people creating accounts, posting content, upvoting each other, and interacting through text on a screen. It took barely five months for that to change. By August 2016, people were already meeting in person — and by 2017–2018, it had exploded into a global phenomenon. The meetup era of Steem was one of the most intense community-building periods in the blockchain's history — and one of the least studied for its lasting effects.
At its peak in early 2018, over 2,670 distinct accounts posted about meetups in a single year. By 2025, that number had fallen to 121. This is the story of the rise, peak, and near-disappearance of in-person gatherings on the chain — and what, if anything, they left behind.
The very first meetups (summer 2016)
Steem was barely five months old when people started meeting in person. The first wave of grassroots gatherings happened in the summer of 2016 — scattered, spontaneous, and often organized by a single person who just wanted to see who else was out there.
On the same weekend in August 2016, two groups of Steemians met for the first time independently of each other. In Siem Reap, Cambodia, @menta organized a Tech Fest meetup on August 13–14 — complete with presentations, custom Steemit t-shirts, and a pizza tuktuk. In Novosibirsk, Russia, @omfedor ran what he called the "first-ever Steemit meetup in Russia" on August 13, posting 18 group photos of attendees. Neither group knew the other was meeting at the same time.
Days later, on August 25, two more meetups happened the same evening. In San Francisco, @ryan-singer hosted 15 Steemians at 20mission with beer and pizza. In Buenos Aires, @sponge-bob documented a gathering of 75+ attendees at the Bitcoin Center — presentations by 
By October 2016, meetups had been documented in at least a dozen cities: @allasyummyfood's London meetup (with Steemit-branded cookies), @jacor's Johannesburg meetup (the first in Africa), @aaronkoenig's Berlin meetup with Ned Scott as special guest, and @dragosroua's Bucharest meetup. Each city's "first Steemit meetup" became a small event in itself — proof that this thing was real, that there were actual humans behind the usernames.
None of this was coordinated. There was no central meetup programme, no funding, no shared playbook. People just wanted to meet each other. That impulse — and the on-chain posts it generated — was the seed of everything that followed.
SteemFest: the flagship (November 2016)
The first large-scale, organized gathering of Steem users was SteemFest, organized by witness 
The event generated 329 posts from 142 distinct authors in its month alone. For a network that had existed for barely eight months, having over a hundred users write about the same physical event was remarkable. The on-chain footprint of SteemFest 1 was larger than the entire meetup-post output of the preceding five months combined.
SteemFest would continue annually, growing each year:
| Edition | Date | Location | Attendance | Countries | Posts (month) | Authors |
|---|---|---|---|---|---|---|
| SteemFest 1 | Nov 2016 | Amsterdam | ~206 | 31 | 329 | 142 |
| SteemFest 2 | Nov 2017 | Lisbon | ~295 | 38 | 1,298 | 407 |
| SteemFest 3 | Nov 2018 | Kraków | 300+ | 47 | 1,260 | 421 |
| SteemFest 4 | Nov 2019 | Bangkok | ? | ? | 502 | 185 |
SteemFest 2 and 3 each generated over 1,200 posts — an extraordinary on-chain footprint for a single event.
But SteemFest was always a conference — a gathering of the already-committed. The story of meetups as an onboarding tool is a different one, and it starts with what happened at the grassroots level.
The grassroots scale up (early 2017)
The summer 2016 meetups proved people wanted to meet. But they were one-off events — a city's "first Steemit meetup" with no second one planned. Through early 2017, the pattern shifted from firsts to repeats:
@eroche— "Meetup in Dublin" (January 2017) — a handful of Irish Steemians in a pub@timsaid— "Steemit Meetup Hamburg" (February 2017) — German community forming@lrock— "STEEM Berlin Meetup" (January 2017)@firepower— "Successful #1 India Steemit Accelerator" (June 2017) — India's first organized meetup
In 2016, 143 distinct authors posted about meetups across the entire year. By mid-2017, that number was climbing fast — 867 for the full year — but the activity was still mostly one-off events organized by individuals. There was no systematic programme, no funding structure, and no repeatable process.
That changed in the second half of 2017.
The Cervantes programme: meetups as a machine (August 2017 – mid 2018)
The first organized, systematic meetup programme on the chain was run by @cervantes, the Spanish-language witness and community project. Starting in August 2017, Cervantes began running meetups across Venezuelan cities at a pace that had no equivalent anywhere else on Steem.
The first was in Barquisimeto: @cervantes — "Proyecto Cervantes Meetup. Primera Conferencia pública en Barquisimeto, Venezuela" (August 4, 2017). Within weeks, the programme had expanded to Maracay, Maracaibo, and Barcelona (Anzoátegui). By the end of 2017, Cervantes had run meetups in at least twelve Venezuelan cities: Barquisimeto, Maracay, Maracaibo, Barcelona, Villa de Cura, Caracas, San Joaquín, Puerto la Cruz, San Francisco, Mérida, Margarita, and Lechería.
Each meetup followed a repeatable format: an announcement post ("Presentación"), the event itself, and a recap post ("Recapitulación") with photos and attendee lists. 
The social footprint was substantial. Each Cervantes meetup generated not just the official recap but a cascade of attendee posts — people writing about their experience, posting selfies with other Steemians, explaining what they'd learned. A single Cervantes meetup in Barcelona (Anzoátegui) in September 2017 produced at least 15 attendee blog posts from individual accounts like 
Running alongside the Cervantes programme, @rutablockchain (co-led by 
The global explosion (late 2017 – February 2018)
The Cervantes programme was the most organized, but it wasn't alone. Across the world, Steem communities began meeting in person — and the pace accelerated dramatically through late 2017 and into early 2018.
Korean meetups emerged in June 2017 as a well-organized, funded programme. 
Indonesian meetups (the KSI — Komunitas Steemit Indonesia) launched in mid-2017, with particularly strong activity in Aceh province. The "Steemit Happiness 2018" event in Banda Aceh was one of the largest organized gatherings outside of SteemFest.
Nigerian meetups started appearing in late 2017 — 
The Promo-Steem movement, championed by 
promo-steem tag, and earn rewards for their promotional efforts. The movement exploded: from 5 promo-steem meetup posts in September 2017 to 293 in February 2018 alone — then collapsed nearly as fast, dropping to single digits by late 2018.
German and Austrian Steemians ran regular "Stammtisch" (pub meetup) events, with 
The numbers tell the story of the explosion:
| Period | Meetup posts/month | Distinct authors (year) |
|---|---|---|
| 2016 | 8–92 | 143 |
| Jan–May 2017 | 13–29 | — |
| Jun–Dec 2017 | 129–354 | 867 |
| Jan 2018 | 925 | — |
| Feb 2018 | 1,394 | — |
| Mar 2018 | 1,010 | — |
| Full year 2018 | — | 2,670 |
(Post counts exclude 
February 2018 was the all-time peak: 1,394 meetup-titled posts in a single month from hundreds of authors across dozens of countries. For context, that month the token price was near its all-time high (STEEM briefly touched $7 in January 2018). Everyone was meeting everyone.
By region, the 2018 meetup landscape looked like this:
This was a genuinely global phenomenon. In 2018, people were meeting in person to talk about Steem in at least 20+ countries across every inhabited continent.
The crash (mid 2018 – 2019)
It didn't last. The crypto winter of 2018 hit meetup activity harder and faster than it hit most other on-chain metrics.
Monthly meetup posts fell from 1,394 in February 2018 to 161 by September 2018 — an 88% decline in seven months. The distinct-author count collapsed from 2,670 (full year 2018) to 460 (full year 2019). Every regional programme contracted:
| Region | 2018 posts | 2019 posts | Change |
|---|---|---|---|
| Hispanic | 444 | 61 | −86% |
| Promo-Steem | 744 | 19 | −97% |
| Korean | 264 | 188 | −29% |
| German/Austrian | 197 | 45 | −77% |
| Indonesian | 141 | 1 | −99% |
| Japanese | 139 | 25 | −82% |
| Nigerian | 78 | 3 | −96% |
The Korean community was the notable outlier — their meetup activity declined only 29%, suggesting a community infrastructure that could survive without bull-market economics. The Indonesian and Nigerian communities effectively disappeared overnight.
The Cervantes meetup programme wound down. The economics that had funded it — STEEM rewards worth real money in Venezuelan bolívars — collapsed with the token price. Rutablockchain continued its online curation and verification work, but the in-person events largely stopped.
SteemFest 4 in Bangkok (November 2019) went ahead, but the on-chain signature told the story: 502 posts from 185 authors, compared to SteemFest 3's 1,260 posts from 421 authors a year earlier. The committed core was still meeting, but the broader community had retreated to their screens.
The double disruption (2020)
Two things happened in quick succession that killed what remained of meetup culture on the chain.
In March 2020, the hostile takeover of Steem by Justin Sun led to the community fork that created Hive. The fork split the community, fragmented social bonds, and forced every project and individual to choose sides. Cervantes, Rutablockchain, and most of the Latin American community moved to Hive. The Korean community largely stayed on Steem (where 
Days later, COVID-19 lockdowns began worldwide. In-person meetups became impossible.
The combination was devastating. Meetup posts dropped to 14–71 per month through 2020, with only 149 distinct authors for the entire year.
The slow revival (2022–present)
In-person meetups returned, but at a fraction of their former scale.
HiveFest 7 (September 2022, Amsterdam) was the first physical gathering since the fork and the pandemic. It generated 275 posts from 116 authors — roughly a fifth of SteemFest 2's on-chain footprint.
More interestingly for the onboarding question, Venezuelan communities restarted grassroots meetups in mid-2022 — independently of HiveFest and without the Cervantes programme's centralized organization. In July 2022, the Hive Monagas meetup in Maturín drew 29 attendees and generated a burst of enthusiastic posts. The same month, Maracay held its first Hive meetup since the pre-fork era, with at least 15 attendee posts. These were organic, community-driven events — not funded by a large witness project, but organized by local Hivers who wanted to meet face-to-face.
In the Philippines, the Cebu teachers community showed what small-scale, workplace-based meetups could look like. In 2023, @indayclara organized a meetup of teachers from Ilihan Integrated School — all onboarded through the OCD programme — demonstrating the same pipeline pattern that had worked for Cervantes years earlier: an existing Hiver onboards colleagues, the group meets in person, and the social bonds reinforce continued activity.
By late 2022, meetup posts had climbed back to 40–80 per month — up from the 2020 nadir of 14–20, but still just 5% of the February 2018 peak.
The overall trajectory:
| Year | Meetup posts | Distinct authors | SteemFest/HiveFest posts | SF/HF authors |
|---|---|---|---|---|
| 2016 | 241 | 143 | 628 | 224 |
| 2017 | 1,744 | 867 | 2,639 | 821 |
| 2018 | 6,198 | 2,670 | 2,397 | 671 |
| 2019 | 1,239 | 460 | 1,374 | 355 |
| 2020 | 322 | 149 | 321 | 158 |
| 2021 | 291 | 115 | 253 | 151 |
| 2022 | 421 | 230 | 480 | 190 |
| 2023 | 410 | 201 | 423 | 146 |
| 2024 | 280 | 135 | 402 | 122 |
| 2025 | 213 | 121 | 209 | 80 |
From 2,670 distinct meetup authors in 2018 to 121 in 2025 — a 95% decline. The question this post turns to next is whether that decline left a measurable mark on the thing that matters most for Hive's health: user retention.
Part 2 — Did meetups affect retention?
The history above tells us that meetups were big, global, and emotionally important. But did they actually move the retention needle? To answer that, we need to separate three things: the people who participated in meetups, the communities that hosted them, and the chain as a whole.
Identifying meetup-associated users
We can't check attendance lists — those don't exist on-chain. But we can identify meetup-associated users through multiple on-chain signals:
- Meetup posters — accounts that published posts with "meetup" in the title (our running dataset from Part 1)
- Meetup-mentioned introductions — accounts whose
introduceyourselfposts mention "meetup," "cervantes," "rutablockchain," "steemfest," or Spanish equivalents like "encuentro" or "conferencia" - Rutablockchain curation trail — accounts whose posts received votes from
@rutablockchain, a direct on-chain record of the users their verification pipeline supported - Cervantes/organizer-welcomed users — accounts whose intro posts received a reply from
@cervantes,@rutablockchain,@danielvehe, or@enmy - Category signals — accounts that posted in meetup-specific categories like
kr-meetupor wrote about meetups underpromo-steem - Cervantes meetup post @-mentions — 95 Cervantes meetup and conference posts listed attendees by
in the body text; extracting these yields 539 verified real accounts
- Cervantes meetup commenters — accounts that commented on
Each proxy captures a different slice of the meetup ecosystem. Meetup posters are the organizers and enthusiastic attendees. Intro-mentioners are people who joined because of a meetup. The Rutablockchain trail data is the strongest pipeline signal — their curation trail voted on 808 unique accounts during their active period, closely matching
Deduplicating across all signals, we can identify 8,126 unique accounts that left an on-chain meetup trace in 2017–2018. That's 1.8% of the 458,510 accounts active in the period.
This is a verifiable lower bound. It captures only the people who left a written trace — a post, a comment, a vote, or an @-mention. Many meetup attendees never appeared in any of these records. The first Cervantes meetup in Barquisimeto drew 27 attendees; the @-mentions in the recap post name 6 of them. If that ratio holds across all meetups — roughly a quarter of attendees appearing on-chain — the true reach was likely three to four times higher, putting it at roughly 25,000–30,000 people, or 5–7% of active users.
Even at the upper bound, meetups touched a small minority. The question is whether that minority retained differently.
The headline numbers
They did — dramatically.
Meetup authors vs. equally-active non-meetup authors (2017–2018)
The fairest test controls for activity level. If someone posted about meetups, they were clearly an active user — so comparing them to all accounts would conflate the meetup signal with a simple activity signal. Instead, we compare meetup authors to non-meetup authors who were equally active: both groups posted at least 5 times in their first 3 months.
| Cohort year | Meetup authors | Retention (6-month) | Non-meetup active | Retention (6-month) | Advantage |
|---|---|---|---|---|---|
| 2017 | 774 | 72.4% | 79,300 | 38.8% | 1.87× |
| 2018 | 787 | 43.8% | 114,323 | 22.7% | 1.93× |
Even after matching on activity level, meetup authors retained at nearly twice the rate of comparable users. This held in both the bull market (2017) and the early bear market (2018).
Intro posts mentioning meetups
Users whose intro posts mentioned meetup-related terms retained at 43.5% (255 of 586), compared to 30.7% for the 74,547 intro authors who didn't mention meetups — a 1.42× advantage. These are people who said, at the moment of joining, that a meetup was part of why they were there.
Rutablockchain's curation pipeline
We can identify the accounts that Rutablockchain's trail actually voted on during their active period (October 2017 – mid 2019) — a direct on-chain record of the people their pipeline supported. This captures 808 unique accounts, closely matching
| Cohort | Total | 6-month retention | Still active 2024–2025 |
|---|---|---|---|
| Ruta-curated | 808 | 79.3% | 175 (21.7%) |
| Hispanic intros, not Ruta-curated | 7,875 | 34.9% | 203 (2.6%) |
Rutablockchain's pipeline retained users at 2.3× the rate of comparable Hispanic newcomers who didn't go through it. And the long-term survival gap is even wider: 21.7% vs. 2.6% — an 8.4× advantage still visible seven years later. Of the 808 people Rutablockchain's trail curated, 175 are still posting on Hive today.
SteemFest/HiveFest authors
People who posted about the flagship conferences showed extraordinary retention:
| First SF/HF year | Authors | 1-year retention | Still active 2024–2025 |
|---|---|---|---|
| 2016 (SF1) | 231 | 77.5% | 38 (16.5%) |
| 2017 (SF2) | 756 | 57.9% | 113 (14.9%) |
| 2018 (SF3) | 555 | 60.9% | 148 (26.7%) |
| 2019 (SF4) | 274 | 66.1% | 108 (39.4%) |
| 2020 (HF5, VR) | 115 | 80.9% | 78 (67.8%) |
| 2021 (HF6, VR) | 93 | 82.8% | 72 (77.4%) |
| 2022 (HF7) | 127 | 75.6% | 87 (68.5%) |
| 2023 (HF8) | 86 | 73.3% | 71 (82.6%) |
The later cohorts are almost entirely composed of long-term survivors. Of the 86 people who first posted about HiveFest in 2023, 71 are still active in 2024–2025 — an 82.6% long-term survival rate. SteemFest/HiveFest didn't just retain people for a year; it marked the most committed members of the ecosystem.
The long-term survivors
The most striking finding is what happened years later. Using our broadest on-chain identification, 661 of 7,430 meetup-touched accounts from 2017–2018 were still active in 2024–2025 — an 8.9% long-term survival rate across seven years and two chain-splitting crises.
Compare that to the 451,080 untouched accounts active in the same period: only 2,922 were still posting in 2024–2025 — a 0.65% survival rate.
Meetup-associated users were 14 times more likely to still be on the chain seven years later.
But here's the number that reframes what meetups meant for the chain: of 3,583 total survivors from the 2017–2018 era who are still active today, 661 are meetup-touched — that's 18.4% of all long-term survivors, from a group that was just 1.6% of the original population. And this is the lower bound: if the true meetup reach was 15,000–20,000 people, the fraction of survivors with a meetup connection could be substantially higher.
Meetups didn't retain users in bulk. They produced a disproportionate share of the people who are still here.
The community-level picture
Individual meetup participants retained dramatically better. But did meetup-heavy communities retain better than communities without meetups?
During the peak meetup era (June 2017 – June 2018), we can compare the Hispanic community — which had the most organized meetup programmes (Cervantes, Rutablockchain) — to the general population:
| Era | Hispanic retention | General retention | Hispanic advantage |
|---|---|---|---|
| Pre-meetup (Jan–Jul 2017) | 50.8% (n=2,204) | 12.5% (n=165,103) | 4.06× |
| During meetups (Aug 2017 – Jun 2018) | 28.7% (n=15,886) | 7.7% (n=756,890) | 3.72× |
| Post-meetup (Jul 2018 – Jun 2019) | 18.0% (n=5,128) | 4.6% (n=220,871) | 3.92× |
The Hispanic advantage is roughly constant at 3.7–4.1× across all three eras. It doesn't widen during the meetup period and doesn't narrow after meetups stop.
This is the most important finding in the analysis, and it cuts against the simple narrative. If meetups were the primary cause of the Hispanic community's retention advantage, we'd expect the advantage to be largest during the meetup era and to shrink afterward. Instead, the advantage is stable.
What this suggests: the Hispanic community's retention edge was driven by something structural — the language-based community, the curation programmes, the social bonds, the community identity — not by meetups alone. Meetups were one manifestation of that community infrastructure, not its sole cause. When meetups stopped, the underlying community infrastructure kept working.
The Korean community tells a similar story. During the meetup era, Korean accounts retained at 26.5% — 3.2× the general rate of 8.2%. Korean meetups were well-organized and persisted longer than most others (declining only 29% from 2018 to 2019 while most regions collapsed by 80–99%). But the Korean retention advantage, like the Hispanic one, appears to predate and outlast the meetup programmes.
Disentangling cause and effect
The data makes two things clear:
1. At the individual level, meetup participation is associated with dramatically higher retention. Meetup authors retain at 2× the rate of equally-active users. Long-term, they're 16× more likely to survive. This is the strongest signal in the dataset.
2. At the community level, meetups don't explain the retention gap. Communities with organized meetup programmes (Hispanic, Korean) retained better than the general population, but their advantage was stable before, during, and after the meetup era.
How do we reconcile these? The most likely explanation is selection plus reinforcement:
- The people who attended meetups were already more likely to stay — they self-selected into an activity that required physical effort and social commitment. This is the selection effect.
- The meetups then reinforced that commitment — putting a face to a username, creating obligations to show up again, forming friendships that made leaving the chain feel like leaving people. This is the treatment effect.
- The community infrastructure that meetups helped build — curation trails, local leadership, onboarding pipelines, group identity — persisted after meetups stopped. This is the lasting effect.
We can't cleanly separate selection from treatment with on-chain data alone. But the fact that the Hispanic retention advantage didn't grow during the meetup era suggests that meetups were more of a catalyst for community formation than a standalone retention mechanism. They helped create the community infrastructure that drove retention, but the infrastructure did the heavy lifting.
Part 3 — What Hive can learn
The meetup era on Steem/Hive was, by any measure, a remarkable period of community building. From 143 meetup authors in 2016 to 2,670 in 2018, from a handful of pub gatherings to organized programmes spanning dozens of countries on every inhabited continent. And the data shows it left a mark: the people who participated are, seven years later, 16 times more likely to still be here.
But the data also tells us what meetups didn't do.
Meetups were never a mass tool
Across all on-chain signals — title keywords, intro body text, category tags, curation trail votes, organizer replies, attendee @-mentions, meetup post comments — we can identify 8,126 meetup-touched accounts in 2017–2018, or 1.8% of active users. The Barquisimeto meetup data suggests only about a quarter of physical attendees leave an on-chain trace, putting the upper-bound estimate at roughly 25,000–30,000 people, or 5–7%.
The 95% decline in meetup activity between 2018 and 2025 is dramatic, but even at peak, meetups were touching a small minority of the user base. They couldn't have been the primary retention mechanism because they never reached enough people. But they punched far above their weight: our identifiable 1.8% produced 18.4% of all long-term survivors still active seven years later.
What meetups actually built: community leadership
Where meetups were powerful was in creating and reinforcing the community core. The 348 meetup-era authors still active in 2024–2025 aren't random survivors — they are disproportionately witnesses, community leaders, curation operators, and event organizers. 
Meetups didn't retain users in bulk. They forged the leadership layer that then built and maintained the community infrastructure — the curation trails, the onboarding pipelines, the local communities — that retained users at scale.
The real retention engine: community infrastructure
The clearest evidence for this comes from the Hispanic community comparison. The 4× retention advantage over the general population exists in every era — before meetups, during meetups, and after meetups. What's constant across all three eras isn't meetups; it's the community itself: a shared language, dedicated curation, structured onboarding, and a social identity tied to place and culture.
Meetups helped build that infrastructure. Cervantes meetups created the local networks. Rutablockchain turned those networks into a systematic onboarding pipeline. But the infrastructure, once built, ran without meetups.
For Hive in 2025–2026, the implication is this: don't try to replicate the meetup era. Replicate what the meetup era built.
The things that drive retention at scale are:
- Language/region-based communities with their own curation and social identity
- Structured onboarding that turns a new account into a community member (not just a content producer)
- Local leadership — people who feel responsible for their corner of the chain
- Repeated social contact — whether that's meetups, community posts, or comment threads
Meetups are one way to create these things. They're not the only way, and in a post-pandemic, low-token-price world, they may not be the most efficient way. The Cebu teachers community in the Philippines, documented in our community density analysis, built high-density, high-retention community clusters without major meetup programmes — through workplace and professional networks instead.
If you're going to do meetups
The data does suggest some patterns for anyone considering community-building events:
Repeatable format beats one-off events. Cervantes ran meetups every week across twelve cities. The Korean community held regular kr-meetup gatherings with transparent expense reporting. The retention benefit comes from sustained contact, not a single spectacular event.
Combine meetups with onboarding infrastructure. Rutablockchain's model — meetup → verified registration → curation trail → writing classes — turned meetup attendees into active chain participants at 79% six-month retention (vs. 35% for comparable Hispanic newcomers), with 21.7% still active seven years later. The meetup alone isn't enough; the pipeline after the meetup is what converts attendance into activity.
Fund structure, not spectacle. SteemFest was glorious, but its retention benefit came from reinforcing the already-committed, not from onboarding new users. The Cervantes grassroots meetups, which cost far less per event, reached more new users and fed them into an onboarding pipeline. Both serve a purpose, but for retention ROI, grassroots wins.
The economics have to work. The meetup era coincided with STEEM at $1–7. Cervantes could fund weekly meetups across Venezuela because the rewards were worth real money. When the price collapsed, the programme wound down — not because the community stopped caring, but because the economics stopped working. Any revival of community-building investment needs a funding model that survives bear markets.
Caveats
Selection bias is the elephant in the room. People who attend meetups are self-selected as more committed and social than average users. The 2× retention advantage over equally-active non-meetup users is our best attempt to control for this, but it doesn't fully eliminate selection — someone who attends a meetup at all is displaying a kind of commitment that we can't perfectly measure through post counts alone.
Meetup identification is approximate. We identify meetup-associated users through title keywords ("meetup") and body text mentions ("cervantes," "rutablockchain"). This captures the subset of meetup participants who wrote about it, not the full set of attendees. Many meetup attendees never posted about the meetup itself, so our "meetup author" group is a subset — likely the more active and engaged subset — of all actual meetup participants.
Community classification is rough. We assign accounts to the "Hispanic" community based on their posting in Spanish-language categories within their first 3 months. This misses bilingual users, users who posted in Spanish later, and users in Spanish-speaking communities that used English-language categories.
Retention windows vary. We use 6-month retention (posted 6–12 months after first meetup post or account creation) for most comparisons and 3-month retention for cohort-level analysis. These are consistent within each comparison but not across all tables in this analysis.
Confounding with token price. The meetup era coincided with the crypto bull market. The "during meetup" era (Aug 2017 – Jun 2018) included months when STEEM was at $1–7, while the "post-meetup" era (Jul 2018 – Jun 2019) saw it fall below $0.50. Retention fell for everyone; the question is whether it fell differently for meetup communities, and the data suggests the community advantage was stable.
Small sample sizes for some groups. The meetup-mentioning intro cohort (586 users) is small enough that the retention percentages have wide confidence intervals. The Rutablockchain curation trail (808 users) and meetup-title authors (3,537 across 2017–2018) are more robust. The directional finding is consistent across all proxies, which increases confidence, but individual numbers should be treated as approximate.
Survivorship in the chain fork. The March 2020 Steem→Hive fork means that users who moved to Hive appear in our retention data, but users who stayed on Steem or left entirely do not. The Korean community, which largely stayed on Steem, is partially invisible in post-fork Hive data. Our long-term survival numbers reflect retention on Hive, not necessarily retention in blockchain social media.
Data: HiveSQL, queried May 2026. Meetup identification via title LIKE '%meetup%' on the Comments table (depth = 0), excluding known noise accounts. SteemFest/HiveFest identification via title matching. All analysis code available on request. For the full retention series, see
Corrections
- The Vienna Stammtisch is organized by