Public Conversation on Proposed Online Platform Legislation: Summary and Toplines
Prepared by Thinkwell
Download the PDF Letter size, for print and sharing.
N Size: 243 people who took a side · Likely range: 50% to 65% · August to October 2026
The open conversation
N Size: 243 people who took a sideThreads: 140Likely range: 50% to 65%August to October 2026
The sponsor’s poll, for comparison
N Size: 1,999 Registered VotersMargin of Error: ± 2%September 10, 2026
Key Findings
People in the open conversation backed the restrictions twice as often as the poll said.
- Among poll respondents who picked a side, 27% backed the restrictions. Among the 243 people in the open conversation who took a side, 57% did, most likely between 50% and 65%. (T1, T2)
- Asked at the end, 45% opposed the laws and 32% supported them. (the sponsor’s poll, final question)
Most of the poll’s questions were worded to favor its sponsor.
- 19 of the poll’s 22 questions were worded toward the sponsor’s side. Where the wording leaned, the conversation differed by 26 points; where it was neutral, by about 7 points the other way. (T3, T4)
- Two blind coders agreed on all 22. (T3)
A blind second check agreed with Thinkwell’s reading 95% of the time.
- A blind second reading agreed with Thinkwell’s judge on 95% of checked pairs, and an independent AI model agreed on 92%. (T5)
- Agree and disagree were never swapped. (T5)
No single thread drives the result: the busiest holds 7.8% of the people.
- 243 people took a side across 140 threads. The busiest thread held 7.8% of them, and the five busiest 22% together. (T6)
Comments posted in a thread’s first 15 minutes were 3 times as likely to reach its top quarter as those posted 3 to 12 hours in.
- Top-level comments posted in a thread’s first 15 minutes reached its top quarter 69% of the time, against 22% for those posted 3 to 12 hours in. (T7)
Method
This research was conducted by Thinkwell. Thinkwell read public social media threads and news coverage about the same laws, from August to October 2026. Nobody was asked anything. Thinkwell found the people who took a side in their own words, matched what they said to the poll’s 22 questions about the laws, and checked those readings blind. 243 people took a side, across 140 threads. The read took 45 minutes. In September 2026 a tech industry group paid a national polling firm to survey registered voters online: 4,103 people about how they use online services, then 1,999 people about the proposed laws. Each survey has a margin of error of 2 points.
The two do not measure the same group, so neither number is the truth on its own. The tables that follow show why they differ, and how far to trust each.
Toplines
Topline Report
N Size: 243 people who took a sideLikely range: 50% to 65%August to October 2026
| Question | Response | Frequency | Percentage |
|---|---|---|---|
| T1 | Of the people in the open conversation who took a side on the proposed laws, which side were they on? | ||
| On the side of the restrictions | 138 | 57% | |
| Against the restrictions | 105 | 43% | |
| Total | 243 | 100% | |
The share on the side of the restrictions most likely lies between 50% to 65%.
Topline Report
Poll: N Size 1,999 Registered Voters, ± 2%Conversation: N Size 243 people
| Question | Response | Frequency | Percentage |
|---|---|---|---|
| T2 | Share on the side of the restrictions: the sponsor’s poll and the open conversation, among those who picked a side. | ||
| The poll, averaged over its 22 questions about the laws | Not reported | 27% | |
| The open conversation | 138 | 57% | |
| The open conversation, likely range | Not reported | 50% to 65% | |
| Difference, conversation minus poll | Not reported | +30 pts | |
Topline Report
N Size: 22 poll questions
| Question | Response | Frequency | Percentage |
|---|---|---|---|
| T3 | How was each of the poll’s 22 questions about the laws worded? Two coders read each one blind. | ||
| Worded toward the sponsor’s side | 19 | 86% | |
| Neutral wording | 3 | 14% | |
| Total | 22 | 100% | |
Two blind coders agreed on all 22.
Topline Report
N Size: 22 poll questions
| Question | Wording | Questions | Gap (likely range) |
|---|---|---|---|
| T4 | Conversation minus poll, by how the question was worded. Points between the conversation and the poll, with the likely range. | ||
| Worded toward the sponsor’s side | 19 | +26 pts (+16 pts to +36 pts) | |
| Neutral wording | 3 | −7 pts (−24 pts to +17 pts) | |
Only 3 questions were neutral, so that row is a first look, not a settled number.
Topline Report
N Size: 161 pairs read twice
| Question | Response | Frequency | Percentage |
|---|---|---|---|
| T5 | A second reader read the same comment and question pairs again, blind. How did the second reading compare with the first? | ||
| Matched the first reading | 153 | 95.0% | |
| A side read as unrelated | 8 | 5.0% | |
| Agree swapped with disagree | 0 | 0.0% | |
| Total | 161 | 100% | |
An independent AI model, reading the pairs blind, agreed on 92% of them.
Topline Report
N Size: 243 people who took a sideThreads: 140
| Question | Thread | Frequency | Percentage |
|---|---|---|---|
| T6 | People who took a side, in the five busiest threads, and in the other 135 threads. | ||
| 1st busiest thread | 19 | 7.8% | |
| 2nd busiest thread | 14 | 5.8% | |
| 3rd busiest thread | 7 | 2.9% | |
| 4th busiest thread | 7 | 2.9% | |
| 5th busiest thread | 6 | 2.5% | |
| The five busiest together | 53 | 21.8% | |
| The other 135 threads | 190 | 78.2% | |
| Total | 243 | 100% | |
Topline Report
N Size: 5,444 top-level commentsAugust to October 2026
| Question | Posted | Comments | Reached the top quarter |
|---|---|---|---|
| T7 | Top-level comments that reached their thread’s top quarter, by when they were posted. | ||
| First 15 min | 747 | 516 (69%) | |
| 15 min to 1 h | 1,148 | 579 (50%) | |
| 1 h to 3 h | 1,254 | 363 (29%) | |
| 3 h to 12 h | 1,408 | 313 (22%) | |
| 12 h to 1 day | 504 | 163 (32%) | |
| 1 day to 3 days | 268 | 106 (40%) | |
| 3 days or more | 115 | 37 (32%) | |
| Total | 5,444 | 2,077 (38%) | |
Early comments get the reach; the chance falls through the first 12 hours and recovers in slow threads.