Foreword — by 
In my previous post, I showed that retained users predict HIVE price increases and that there's no detectable sell pressure from user growth at the aggregate level. But as I noted at the end of that post, there's a sharper version of the question. 
"Hmmmm, how much are the users earning in these communities and how much is actually being powered up? While retention rates are high, are these users benefiting the economy or just draining it? If they are draining it, is seeking a higher retention among them a good idea? Are the earnings the only reason they are being retained?"
This is a fair question, and one that deserves a data-driven answer rather than impressions. It captures a critique I've encountered many times sometimes stated diplomatically on chain, sometimes much more bluntly on Discord. The argument is that retaining users who sell their rewards might actually be worse for Hive than losing them. That high retention in economically distressed communities isn't a success story but a warning sign.
This is a testable hypothesis. Below are the results.
Research Findings — by Claude Opus 4.6
The following analysis was conducted at
The Critique
This is a serious economic argument, and stating it clearly matters. The weaker version — that these users are lazy, disengaged, or producing low-effort content — is easily tested and, as we'll show below, easily refuted. But the stronger version doesn't depend on effort or engagement. It says: even if they work hard, they're still a net drain, because they sell.
The implication is that retaining these users may be worse for Hive than losing them. That the platform would be better off with fewer users who stake more.
This post tests the critique in two parts. First, the behavioral claims: are these users actually disengaged, low-effort, or short-term? Second, the economic question: even if they're active, does their participation create value for the network — or is it subsidized activity that goes nowhere?
The behavioral data is straightforward. The economic question is harder, and answering it honestly requires looking at both what's worked and what hasn't — including projects where serious fraud allegations emerged. We'll look at all of it.
Part 1: The Behavioral Claims
Before we get to the economic question, let's dispose of the simpler version of the critique — that these users are lazy, disengaged, or here for a quick buck. We tested this across 8,156 accounts created between 2021 and 2023, classified by self-reported location. Venezuelan and Cuban users are split by whether they entered through the organized Spanish-speaking community. USA and UK serve as the affluent-country baseline. (Full tables, charts, and methodology are in the comment below.)
The results are unambiguous:
- They don't leave. Venezuelan+Spanish users retain at 45% at six months and 14% at three years — roughly double the USA rate (28% and 7%) at every time horizon. Cuban+Spanish is even stronger.
- They curate broadly. The median Venezuelan+Spanish user votes for 27 distinct authors; Cuban+Spanish: 123. The median American user: 9. Self-vote rates are lowest in the Spanish-speaking groups (0.2% median).
- They engage. The median Cuban+Spanish user writes 4.8 comments for every blog post. Venezuelan+Spanish: 1.3. USA: 0.2 — one comment for every five posts.
- They produce substantial content. The median Venezuelan+Spanish post is 5,199 characters — nearly 3× the USA median. Only 1% average under 500 characters, versus 20% for USA. (A caveat, flagged by @erikah: many Hispanic users write bilingual posts — Spanish original plus an English translation, often machine-generated. This inflates character counts relative to monolingual posts. Even accounting for this, the near-absence of sub-500-character posts contrasts sharply with the USA cohort, but the raw length gap overstates the difference in original writing effort.)
- They do power down more (37% vs. 16% for USA). This is the one finding that partially supports the critique. But the difference is in amounts, not commitment: Cuban+Spanish users power up at the highest rate of any group (43%), and the average Venezuelan stakes 11 HIVE versus 205 for an American. That's a wealth gap, not a behavior gap.
These users are not freeloading. By every behavioral metric — retention, curation, engagement, content effort — they are among the most active participants on the platform. The typical American user on Hive is less engaged than the typical user from every distressed country tested.
But the stronger version of the critique already concedes this. It says: fine, they work hard — but they still sell, and selling is what matters. That's an economic argument, and it requires an economic answer.
Part 2: The Economic Question
The extraction critique rests on a simple model: rewards go out, users sell, the token price drops. Under this model, the only user worth retaining is one who stakes — and anyone who sells is, at the margin, making every other holder poorer.
This model isn't wrong about the mechanics. Selling does create sell pressure. But it's incomplete in ways that matter.
Inflation happens anyway
Hive's reward pool is funded by protocol-level inflation of roughly 3.5% per year. This inflation occurs regardless of who receives the rewards. If distressed-country users didn't earn them, other users would — and those users also sell. The USA power-down rate in our cohort is 16%, not zero. The relevant question isn't "do these users sell?" but "does Hive get more value from allocating rewards to active communities than it would from the alternative allocation?"
The alternative is not "no inflation." It's the same inflation distributed among fewer, less active users.
What gives a token value?
The extraction model treats HIVE as a pure speculative asset — something whose value comes from people holding it and not selling. But no token can sustain value on holding alone. Long-term, a token is worth what the network underneath it is worth. And a network's value comes from people using it.
This means there are three sources of demand for HIVE:
- Speculative demand — people buying because they expect the price to rise. This is what the extraction critique focuses on exclusively.
- Utility demand — people needing the token to do things: Resource Credits for transactions, HIVE Power for curation influence, HBD for payments.
- Network-effect demand — a platform with active users, content, and communities attracts more users, developers, and investors. This is indirect, but it's what sustains every social platform's valuation.
The critique says distressed-country users don't contribute to (1). That's largely true. But it ignores (2) and (3) entirely — and those are the sources of demand that scale.
A blockchain where a few hundred whales hold tokens and nobody posts is not valuable. A blockchain where thousands of people post daily, curate each other's work, and spend the native stablecoin at real businesses has a product. The question is whether the activity funded by inflation is building something — or just burning money.
The case for network effects
Why would network effects develop here, in economically distressed communities, rather than in affluent countries?
Because crypto solves a real problem in these places. In the USA, Hive competes with Instagram, YouTube, Substack, Patreon, and a functioning banking system. The value proposition is weak. In Venezuela, Hive offers something that doesn't exist locally: a censorship-resistant publishing platform with a built-in payment system denominated in a stable currency, accessible to anyone with an internet connection, in a country where the local currency has been destroyed by hyperinflation and the banking system is unreliable.
This isn't theoretical. The behavioral data from Part 1 already shows the result: Venezuelan and Cuban users retain at nearly double the rate of Americans. They engage more, curate more, and produce more content. The platform is more useful to them, and they behave accordingly.
If Hive is going to develop sustainable network effects anywhere, it will be where the platform solves a real problem. The communities most criticized for extraction are the ones where product-market fit is strongest.
The counterfactual
Before looking at the data, consider a thought experiment. Imagine Hive took the extraction critique to its logical conclusion and stopped rewarding users who sell. What would that look like?
Based on Part 1 alone, the platform would lose its most active content creators, its most engaged commenters, and its most prolific curators. The remaining users — mostly affluent-country holders who post infrequently and vote for 9 distinct authors — would be staking more, but staking a token for a platform nobody uses.
Hive's inflation is going to be spent regardless. The question is whether it's invested in building a network that people actually use, or distributed to passive holders of a token with declining utility.
But maybe that trade-off is worth it — if these communities really are just selling and contributing nothing beyond content. The data below tests whether that's true.
The HBD economy
The strongest concrete evidence for utility demand is HBD — Hive's dollar-pegged stablecoin — being used to buy real goods and services. When a user earns HBD and sends it to Binance, that's sell pressure. When the same user spends it at a grocery shop, sends it to family, repays a loan, or pays for a service, it's not. The HBD stays on-chain, demonstrates utility, and builds the habits and infrastructure that give a stablecoin actual value.
The extraction critique assumes most HBD goes straight to exchanges. We tested this by analyzing every HBD transfer on the blockchain from 2022 through May 2026. To isolate real economic activity, we filtered out over 120 known infrastructure accounts — exchanges, swap services, curation bots, tip bots, game accounts, DAO transfers, and platform services. We excluded self-transfers and capped transfers at 500 HBD to focus on grassroots activity rather than whale movements. What remains is person-to-person and person-to-merchant transfers: the real HBD economy.
The scale is far larger than any individual project might suggest:
| Year | Transactions | Unique senders | Unique receivers | HBD volume |
|---|---|---|---|---|
| 2022 | 51,828 | 8,082 | 5,965 | $637,448 |
| 2023 | 43,148 | 6,162 | 4,704 | $671,294 |
| 2024 | 57,110 | 6,989 | 4,790 | $823,457 |
| 2025 | 62,153 | 7,615 | 5,482 | $797,453 |
| 2026* | 8,234 | 1,726 | 1,228 | $105,260 |
*January–May only. HBD printing from author rewards stopped in February 2026; existing HBD continued to circulate.
Over four years, this economy processed $3.0 million through 222,000 transactions involving thousands of unique participants each year. The average transfer was $12–16 — consistent with real commerce, not whale movements.
What does this economy look like? We sampled hundreds of individual transfers and examined the memos, accounts, and profiles involved. The activity spans at least four countries and multiple categories:
- Venezuela: Purchases at shops via Keychain Store and V4V payment tools — bakeries, supermarkets, restaurants. Savings accounts, health funds, and aid for medical emergencies.
- Cuba: A P2P exchange service (
) converting HBD to local currency, a dedicated remittance service (
— "Remesas a Cuba con HIVE y HBD, 10min–48hrs"), and a currency exchange service (
).
- Nigeria: Grocery shops (
in Aba — "Major groceries shop"), household goods merchants (
— "Sell all household items and beverages"), P2P crypto-to-fiat services (
), and community workshop funding.
- Cross-border: A lending service (
) issuing numbered loans and receiving installment repayments, service payments, and personal transfers between family members.
A sample of real transfers from the blockchain illustrates the variety:
This is not a single organized project. It is an organic, multi-country economy that grew from the bottom up — users discovering that HBD works as money and building the habits and services to use it that way.
How much did these communities actually extract?
The extraction critique makes a specific accusation: that Hispanic communities drain value from Hive by selling their rewards. We can test this directly by measuring how much HBD each cohort group actually converted out — through every available pathway: direct exchange deposits, the internal market (Hive's built-in DEX), and protocol conversions.
But raw totals are misleading, because the Hispanic cohort is larger and has higher retention. A fair test requires comparing like with like. We sampled 408 Hispanic users still posting in 2025 against 2,000 randomly sampled active users from the same 2021–2023 account creation window — same tenure, same activity requirement. The result:
Posts per user: 1.02×. Voting breadth: 1.01×. HBD disposed per user: 1.01×. On a properly controlled comparison, Hispanic active users are statistically indistinguishable from their peers — on both engagement and extraction.
In absolute terms, the Hispanic cohort's total HBD disposal across all pathways was $142,000 in 2025 — roughly 1.1% of the $12.8 million converted network-wide in 2025. The extraction problem, to the extent one exists, lies elsewhere entirely.
What killed the HBD economy
Through late 2025, the real economy began declining — not because users lost interest, but because HBD was becoming scarce. As HIVE's price fell, the dollar value of author rewards dropped, and less HBD flowed into the communities that were spending it.
Then in February 2026, Hive's protocol delivered the final blow. The blockchain includes a debt limit: when the total HBD supply exceeds 20% of HIVE's market cap, HBD printing from author rewards stops entirely. On February 3, 2026, the debt ratio crossed 20%. HBD payouts dropped to zero by March. All author rewards are now paid in HIVE and HP only.
The effect on the real economy was immediate and devastating:
| Month | Real economy tx | Real economy HBD | Exchange tx | Exchange HBD |
|---|---|---|---|---|
| 2025-10 | 4,945 | $67,919 | 2,891 | $142,058 |
| 2025-11 | 3,867 | $48,735 | 2,275 | $216,817 |
| 2025-12 | 3,555 | $45,206 | 2,047 | $163,388 |
| 2026-01 | 3,046 | $30,913 | 2,031 | $51,767 |
| 2026-02 | 1,920 | $28,240 | 1,206 | $97,840 |
| 2026-03 | 1,375 | $21,098 | 652 | $34,250 |
| 2026-04 | 1,380 | $16,982 | 511 | $23,629 |
| 2026-05 | 513 | $8,028 | 199 | $11,014 |
Both the real economy and extraction collapsed together. But even during the collapse, the real economy consistently generated 2–3× more transactions than exchange deposits. The people using HBD as currency outnumbered the people dumping it on exchanges right up to the end.
The HBD economy didn't fail because the model was broken. It starved.
What this means for the extraction argument
The data doesn't just weaken the extraction critique. It undermines it on its own terms.
The communities accused of draining value from Hive were building the platform's only organic, growing use case for its stablecoin — a $3.0 million, 222,000-transaction economy spanning multiple countries, with merchants, lending services, remittance platforms, and P2P exchange networks.
We should be honest about the limits of this analysis. We cannot precisely quantify the dollar value of the network effects this economy creates. Some of the filtered transfers are personal gifts or tips rather than commerce. And we cannot prove that HBD circulating through a grocery purchase generates more long-term value for the HIVE token than the same HBD going to an exchange. That is partly an economic reasoning argument, not a pure data claim.
But the data does establish several things that the extraction critique cannot dismiss:
- The accused communities account for 1.1% of total network HBD disposal — a rounding error in the macro picture.
- Their disposal-to-earnings ratio is not exceptional compared to other cohort groups.
- Thousands of people in economically distressed countries were using HBD as actual money — for groceries, loans, remittances, services.
If Hive's stablecoin has a future as a medium of exchange — which is the most compelling use case for any stablecoin — it will be built by the communities currently accused of extraction. They are the ones who demonstrated it works.
The governance implication
The HBD debt limit exists for a reason — it prevents the stablecoin from becoming undercollateralized. But the current design creates a brutal trade-off: the same mechanism that protects HBD's peg also destroys HBD's utility as a medium of exchange, precisely when the ecosystem most needs economic activity.
This matters for the extraction debate. The strongest evidence for these communities' economic contribution is the real economy they built — and the protocol has now shut off its currency supply. Every month the debt ratio stays above 20%, the merchant networks, payment habits, lending services, and local economic relationships that took years to build continue to atrophy.
The communities accused of extraction built the closest thing Hive has to a real-world use case for HBD. Whether that use case survives is now a governance question. The debt ratio that triggered the printing halt is driven in part by the market cap of HIVE — which DHF spending directly affects. The viability of HBD as a medium of exchange depends on stakeholder fiscal discipline with the DHF as much as it does on protocol design.
Caveats
Location classification is approximate. We identified countries from self-reported location strings in profile metadata (e.g., "Caracas" → Venezuela). Users who don't fill in their profile, or who report misleading locations, are misclassified. The cohort is 8,156 accounts out of roughly 250,000 created in 2021–2023 — those with identifiable locations and at least one blog post.
Spanish-language tagging is a proxy for community membership. We split Venezuelan and Cuban users by whether their first post used Spanish-language tags (hive-196387, cervantes, spanish, etc.). This distinguishes users who entered through the organized Spanish-speaking community from those who did not. The behavioral differences between these subgroups are large — Venezuelan+Spanish users outperform Venezuelan non-Spanish users on nearly every metric — suggesting the split captures something real, but it's imperfect.
Survival bias in retention figures. The 2yr and 3yr retention rates are restricted to cohort members with enough elapsed time, but the users still posting at those horizons are by definition the most engaged. The country-level comparisons hold at every time horizon; the behavioral metrics (post length, voting diversity, comments per post) are computed across the full cohort, not just survivors.
Voting diversity overcounts across years. We aggregated vote quality data year by year, then summed. An author voted for in both 2021 and 2022 counts as two distinct authors. This inflation is proportional across all groups, so the country-level comparisons are valid, but the absolute "distinct authors voted" numbers are upper bounds.
Body length includes markdown, HTML, and translations. (h/t @erikah) Post length is measured in raw characters, which includes markdown formatting, image embed tags, and other markup. Posts with many images may appear longer than their readable text content. Additionally, many Hispanic users write bilingual posts — the original in Spanish plus an English translation, often machine-generated — which roughly doubles the character count compared to a monolingual post covering the same content. This practice is common in the Spanish-speaking community and means the 3× length gap versus USA posts overstates the difference in original writing effort. The near-absence of sub-500-character posts in the Hispanic cohort (1% vs. 20% for USA) is a more robust indicator of content effort, since even halving a bilingual post's length would keep most well above that threshold.
We cannot measure content quality, only content effort. Long posts with images are not necessarily good posts. The analysis tests the "low-effort" claim by measuring effort, not quality. Quality is subjective and cannot be measured from structured data alone.
The cohort is 2021–2023. These findings describe users who joined during this period. Earlier cohorts (2016–2020) may show different patterns, particularly during the Steemit era when the platform's demographics and incentive structures were different.
Economic context is inferred, not measured. We describe Venezuela, Cuba, Nigeria, the Philippines, and Indonesia as "economically distressed" based on widely available macroeconomic data. We do not have individual-level economic data for any user. A Venezuelan user on Hive may or may not be economically distressed; a US user may or may not be affluent.
Comment balance depends on community visibility. Small, high-profile groups (like UK users) receive more comments from the broader platform relative to their size. The ratio metric controls for this better than the raw difference, but community visibility effects cannot be fully eliminated.
The real economy filter is conservative but imperfect. We excluded over 120 known exchange, swap, bot, tip, game, curation, and platform accounts from HBD transfers to isolate person-to-person and person-to-merchant activity. Some legitimate economic transfers may have been excluded (e.g., a swap intermediary that facilitates real commerce), and some noise remains (personal gifts, contest prizes). The filter was developed by iteratively examining the top senders and receivers and sampling transfer memos. The overall trends are robust to individual account misclassification, but the absolute numbers are approximate.
We cannot fully measure HBD velocity. Tracing a specific unit of HBD through multiple hops (customer → merchant → supplier → next merchant) requires transaction-graph analysis that is beyond the scope of this study. Low merchant balances suggest HBD moves quickly through multiple hands, but we cannot quantify the average number of on-chain hops before exit.
The economic argument is partly theoretical. The case that real economic activity generates more long-term value than the equivalent amount of sell pressure rests on economic reasoning about network effects, not solely on measurable data. We can show that the real economy was substantial and growing while extraction was declining, but we cannot directly measure the dollar value of network effects or prove that activity translates to token demand at a specific rate.
Exchange deposit list was validated but may not be exhaustive. We compiled exchange and swap accounts from known major services, then verified by examining the top receivers by volume in the filtered data. No significant exchange accounts appeared in the residual. Some local P2P exchange services (e.g.,
Data: HiveSQL, queried May 2026. Behavioral cohort: 8,156 accounts created January 2021 – December 2023, classified by profile location metadata. Activity data: January 2021 – May 2026. Voting data: January 2021 – December 2025. HBD transfer data: January 2022 – May 2026, filtered against 120+ known infrastructure accounts. All queries and analysis scripts available upon request. For prior posts in this series, see