Dmitriy Lastovka (@mnemath1) header
Dmitriy Lastovka (@mnemath1) profile picture
Joined September 2019
Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)Photo by Dmitriy Lastovka (@mnemath1)
If you’re here, you probably have something important to say. For our voices to be heard louder, we need to make the news together. @DappExpert and @DeFight_Game are a gamified AI news aggregator for Web3 projects. Let’s build a community and shape our own narrative — different from global media.
3
2
7
260
Four free postcards in a donation letter lifted the share who gave from 12% to 21%. A gift makes many people feel they owe you. Hide the string, and value turns into a debt. #Psychology #Crypto
2
18
Four free postcards in a donation letter: the share who donated rose from 12% to 21%. A gift makes people feel they owe you. Hide the string, and value turns into a debt. Read it on Dapp​.Expert: dapp.expert/analytics/offeri… #Psychology #Crypto
🤖 Made with AI
1
10
One random up-vote made the next reader 32% more likely to up-vote. A crowd is only evidence when it is real and chose independently. Online, it often is neither. #Psychology #Crypto
29
One random up-vote made the next reader 32% more likely to up-vote. Final ratings ended 25% higher. Social proof helps when the crowd is real and independent. Online, it often is neither. Read it on Dapp​.Expert: dapp.expert/analytics/when-s… #Psychology #Crypto
🤖 Made with AI
6
Any tool that turns text into a verdict, from "78% positive" to "written by AI", hides two questions: compared to what, and measured how? Seven questions, tested on our own study. #Research #Writing
12
Before you trust a study of someone's writing style, ask: compared to what, and who checked the labels? Our own study of three writers graded itself B, not A. Two machine coders agreed, but machines can share a bias. Read it on Dapp​.Expert: dapp.expert/analytics/what-m… #Research #Writing
🤖 Made with AI
6
Molly White's verdicts on crypto are harsh, and her texts carry the most numbers, links and footnotes in our study. How she builds criticism the reader can check, step by step. #Writing #Crypto
21
"That would be fraud." Molly White wrote that only after the testimony, the exhibits, her own reading and a tested innocent explanation. Sharp criticism can still be checkable. Write the verdict last. Read it on Dapp​.Expert: dapp.expert/analytics/how-mo… #Writing #Crypto
🤖 Made with AI
9
Benedict Evans compares new tech with old precedents constantly, and says analogies can't predict anything. That's not a contradiction. It's the method. A five-question check for your own comparisons. #AI #Writing
2
40
Benedict Evans compares AI with fibre, mobile data and chips, then writes: "analogies don't have predictive value". A comparison should find the question, name one shared feature and say where it breaks. Not predict. Read it on Dapp​.Expert: dapp.expert/analytics/how-be… #AI #Writing
8
Why Ethan Mollick's AI posts are useful: result first, limits right beside it, advice tied to a situation. What writers can borrow, and the part you can't fake. #AI #Writing
17
Count the words in each sentence of a DeFi explainer. If they all come out near twenty, the reader can't tell which one is the warning. What three analytical writers do instead. #Writing #Editing
1
28
A long sentence carries the chain: loan, collateral, rule, trigger price. A short one lands the point. "Then the protocol can sell your collateral." If every sentence is 20 words, the warning looks like the rest. Read it on Dapp​.Expert: dapp.expert/analytics/senten… #Writing #Editing
6
AI drafts sound alike: when one model suggests to many writers, their texts drift together. Your voice is a few decisions: the opening, what you count, how sure you sound, how you end. Keep those. Delegate the rest, then check. Read it on Dapp​.Expert: dapp.expert/analytics/keepin… #AI #Writing
1
17
"100% sure" is a feeling, not a finding. Next time a post shouts, look for three things: – where the number comes from – how it was measured – when, and over what period If you cannot find them, the confidence is decoration. The calibrated version of a claim (a smaller claim, a date, a link) is less exciting and much more useful. Read it on Dapp.Expert: dapp.expert/analytics/why-a-…
🤖 Made with AI
11
"I tested it for a month." Did you, though? If the testing was done by researchers, say so — and link the study. The information stays the same. What changes is who it belongs to, and whether the reader can check it. Writing about research without borrowing someone else's experience (video narration is AI-generated): dapp.expert/analytics/writin…
🤖 Made with AI
13
Imagine a wrong number goes live: 18% of supply unlocks on 1 March. A reader checks the vesting contract. It is 8%, on 1 April. You can fix it silently. Or you can write one dated line at the top: what we said, what is correct, what it changes. The second version takes a minute and gives the people who already repeated the error a way to learn it was wrong. How to write it, when it is a correction vs an update, and what research says about whether it works: dapp.expert/analytics/correc…
🤖 Made with AI
13
"AI was used" is not a disclosure. It is a shrug. The useful version follows the article through its stages. An example trail: – draft: argument written by a person – translation: AI-assisted wording – check: every fact verified by a person against its source Readers do not need a confession. They need to know who did what, and who answers for the result. Here is a template you can copy: dapp.expert/analytics/explai…
🤖 Made with AI
15
A carousel benchmark can suggest what to test. Check the denominator and whether the formats had comparable assignments. Read it on Dapp.Expert: dapp.expert/analytics/what-c…
🤖 Made with AI
14
A carousel works when each swipe answers a question the previous card leaves open. Start with a Bitcoin payment at zero confirmations. Read it on Dapp.Expert: dapp.expert/analytics/buildi…
🤖 Made with AI
12