This is a print copy. The interactive version, where every chart can be switched between groups, is at dalecaddy.cc
Dale
Caddy
Four years answering the questions DeFi communities actually ask, at the speed they ask them.
The projects I supported
| Project | Chain · type | Peak deposits (TVL) | Peak | Largest chat recorded | My role |
|---|---|---|---|---|---|
| Geist | Fantom · lending | $3.75bn | Oct 2021 | 6,218 members | Main community, wind-down announcements |
| Ellipsis | BNB Chain · exchange | $3.46bn | Apr 2021 | 9,556 members | Main community, set up its Discord |
| Solidly | Fantom · exchange | $2.29bn | Mar 2022 | – | Supported |
| Solidex | Fantom · yield | $2.27bn | Mar 2022 | – | Supported |
| Blizz | Avalanche · lending | $1.30bn | Nov 2021 | 3,359 members | Occasional cover |
| DotDot | BNB Chain · yield and voting | $89.3m | May 2022 | 1,178 members | Time-zone lead, weekly graphics |
| TopShelf | BNB Chain, Fantom, Avalanche · borrowing | $37.7m | Jan 2022 | – | Supported |
TVL is the value deposited in a protocol's contracts, from DefiLlama. These are each protocol's own figures, not funds I managed, and the peaks came at different times, so they don't add up. Chat sizes are the largest member counts in archived copies of each main Telegram group's public page. On trading volume, Solidly handled $2.87bn in its first full month, March 2022, and $5.3bn in total; DefiLlama's Ellipsis volume records start in September 2022, after its busiest period, so they're left out.
Messages per month
Messages still in the group's history, with mine in blue
Show the numbers
Group size over time
Member counts from archived copies of the group's public page · my years there shaded
Show the numbers
How fast I replied
Replies made with the reply button, by wait
Show the numbers
What people asked me
A sample: replies matched by keyword
Show the numbers
When I was there
My messages by hour, Perth time
Show the numbers
Closing Geist after the Multichain collapse
In July 2023 Multichain, the bridge behind many tokens on Fantom, collapsed, and the Geist team decided to close its lending market for good. The decisions and dates were theirs. Telling the community, and helping people get their funds out, was mine.
Geist group, July 2023, day by day
UTC days · my messages in blue
Show the numbers
The shutdown notice
The Website will remain online for at least two more months or so to allow locks to expire, borrows to be repaid and withdraws to happen.The day before, 150 of 364 messages were mine
A firm date, with the contracts listed
The Geist Finance website will be taken down on September 30th.64% of messages in the next 48 hours were mine
Manual withdrawal steps
Which contract, which functions, and the amount people get wrong: 150 GEIST is entered as 150000000000000000000.
Promised in the chat, then deliveredAcross July to September 2023 I wrote 455 of the group's 1,464 messages, replied directly to 80 people, and answered questions in a median of about 3 minutes. A typical explanation: borrows have to be repaid before you can withdraw from fully utilized markets
Vote incentives, every week
I made every weekly graphic ranking the pools that projects paid the most to attract votes to. DotDot called these incentives bribes. Its account published the posts; I led its community in my time zone.
Market worry, answered with facts
When the community worried about EPX's status on Binance, I acknowledged it plainly, then gave current figures for the protocol.
The protocol itself works as intended, has teams actively using it the way it was intended, while also using and bribing via DotDot.
A withdrawal that wouldn't go through
I walked a member through a direct attempt on BscScan, which also failed, and escalated to the developers. Their test transactions found a rounding error in the 3pool contract. I explained their tested workaround, withdrawing the exact balance read from the contract, and drafted an announcement for everyone else.
Criticism is welcome. Breaking rules isn't.
After a ban I'd point out publicly which rule was broken, then often message the member privately: keep it out of the main chat and I'll unban you in a few days. I set up bots, spam filters and quick-answer commands, wrote moderation procedures, supervised moderators, and translated explanations for members who didn't speak English as a first language.
Measured from each group's history
| Group | Period | My messages (at least) | Rank | Days posted | Replied to | Median reply | Largest group |
|---|---|---|---|---|---|---|---|
| Ellipsis | Apr 2021 – Apr 2025 | 9,674 | 2nd of 8,710 | 71% | 1,882 people | ~2 min | 9,556 (May 2022) |
| Ellipsis price chat | Apr 2021 – Apr 2025 | 12,405 | 3rd of 3,302 | 47% | 470 people | ~2 min | 1,920 (Nov 2022) |
| Geist | Oct 2021 – Dec 2023 | 3,560 | 1st of 4,570 | 60% | 805 people | ~1 min | 6,218 (Dec 2021) |
| Geist price chat | Oct 2021 – Sep 2023 | 396 | 27th of 2,031 | 90 days | 71 people | ~4 min | 635 (Jan 2023) |
| DotDot | Apr 2022 – Apr 2025 | 1,781 | 1st of 846 | 33% | 296 people | ~2 min | 1,178 (Apr 2023) |
| Blizz | Feb – Jun 2022 | 42 | Occasional cover | 14 days | 11 people | ~5 min (10 replies) | 3,359 (Dec 2021) |
Every count here is a minimum. Telegram numbers a group's messages in order, and only part of each history survives: 32% of the Ellipsis main group's, 69% of its price chat's, 46% of DotDot's, 51% of Geist's, 72% of Geist's price chat's and 73% of Blizz's. The rest were deleted by members or moderators, removed as spam, or were join notices. My DMs and the Ellipsis Discord aren't counted at all, and the announcement channels were posted to by the developers only. I also sent 15 messages in DotDot's small price chat. Counted from exports of each project's public Telegram group, for my own accounts only. Reply times cover replies I made with Telegram's reply button to messages containing a question, answered within a day: they show how fast I answered when I did, not that every question was answered. Topics are a sample: only questions I answered with the reply button that matched a keyword, and a question can count under more than one. The Geist history covers about half the group's messages, so its shares are approximate. These measure activity, not growth or outcomes. Group sizes are the largest member counts in Internet Archive copies of each group's public page; months without a copy are unknown, and the price chats were saved only a few times, so their peaks were likely higher. My posts appear as Andromeda in Ellipsis, Hera in DotDot and Blizz, and a deleted account in Geist. I also supported Solidly, Solidex and TopShelf, and can ask former Ellipsis and Geist staff to confirm my role.
Looking for remote community management work
Based in Albany, Western Australia (UTC+8), and comfortable working overnight to overlap with teams in other time zones.