Founder of @pulsarplatform, host of @audiencedisco
London, England
- Tweets 16,810
- Following 4,907
- Followers 2,934
- Likes 20,823
Francesco D’Orazio retweeted
EA AI safetyism increasingly looks like Marxism-Leninism for the algorithmic age.
The old vanguard claimed privileged knowledge of the inevitable course of History. The new one claims privileged knowledge of the probabilistic course of Humanity.
Both use an elaborate intellectual framework to reach the same political conclusion: a small group of enlightened people must constrain everyone else for their own good.
That has never lead to anything except monumental human suffering.
275
1,580
9,150
1,164,912
Francesco D’Orazio retweeted
WE'D LIKE TO BUY THE WORLD SOME TIME...
Made with AI
141
384
2,527
271,336
For the past two months we've been running a weekly audience-intelligence read on one of the cleanest natural experiments in public-opinion modeling: Iceland's EU-accession referendum on 29 August. 🇮🇸 📷
Iceland is about as close to a perfect testbed for social intelligence as you'll find:
1) Small, bounded population. You can reach the whole engaged electorate, not a sample.
2) Near-total digital penetration. The usual "social isn't the public" gap is smaller here than almost anywhere.
3) Language as a firewall. Icelandic is spoken almost nowhere else, so "in Icelandic" is very nearly "from Icelanders." (the national signal is remarkably clean, a reminder that data hygiene is the whole game.)
4) Legible influence. In a nation of 380,000, the persuaders are named people, not anonymous swarms. You can watch a single business founder move the entire system in a week.
5) And the part that matters most for modeling is that the experiment ends in a dated, binary, verifiable result. A rare chance to calibrate discourse against reality. So we're publishing the read.
Every week, @PulsarPlatform's Saga — our autonomous research agent — produces a "Dispatch" on how the two camps' narratives form, which are gaining momentum, and where audience sentiment diverges from the press.
Three things the data surfaced this week that a poll wouldn't:
1) The referendum has migrated off the news pages and onto the feed: news is now under 9% of the conversation.
2) Volume and engagement tell different stories. The No frame is being heard roughly 10x louder per author than the other.
3) This week we've started seeing influential voices driving discourse.
@kerecis founder Guðmundur Fertram, endorsed the No side. Clearly discourse is not a vote. Loud is not the same as winning. But as a live lab for what social intelligence can — and can't — tell us about how public opinion forms, it's hard to beat.
Base case, moderate confidence: a No result on 29 August, 55–63%, on a rural + business-elite + identity coalition. The one variable that would compress it: a Fertram-tier Yes-side entrant before the 14 Aug debate. None has appeared.
Read the latest Despatch, link in comments #SocialIntelligence #AudienceIntelligence #PublicOpinion #SocialListening #Iceland #EU
1
125
Kimi K3 by @Kimi_Moonshot smashed through the experience benchmarks climbing to #3 at launch: it's the biggest new arrival since @OpenAI's ChatGPT 5.6, praised for frontend taste, agentic autonomy and debugging without refusals. “Fable refuses, Kimi solves” is the meme of the week, a positioning gift no benchmark buys.
Qwen 3.8 by @Alibaba_Qwen arrives three days after Kimi as a preview, framed almost entirely on economics: cheaper, faster, comparable-enough. It doesn’t own a capability dimension outright, its moat is tokens per dollar, and it holds only while the price holds.
Saga dives into the experience benchmarks in the latest Day One report dayoneai.tiiny.site/
3
29,557
A referendum on the resumption of European Union membership negotiations is due to be held in Iceland on 29 August 2026.
With about 5 weeks to go, the No side is winning on autopilot: it still owns social discourse 2.8-to-1 and frames 8 of the 10 most-visible posts. And yes its momentum is cooling, but the Yes camp added no new credible public voice to its side, and the persuadable middle is publicly asking where the Yes content is.
Saga deep dives into the Yes camp no-show in our latest Dispatch 03👇 lnkd.in/ei6WuW4S
1
2
1,333
Are people starting to go back to @OpenAI form @AnthropicAI ?
Definitely not an exodus yet, but we're starting to see the beginning of an emerging defection narrative forming.
lnkd.in/e777UP2u
1
2
54
Benchmarks show lab wins but real adoption is scattered across thousands of user posts. Day One, powered by Saga, tracks what early users love and hate about every model across 18 dimensions, from launch day onward.
The focus of this first release: @OpenAI ChatGPT 5.6, @SpaceX @xai Grok 4.5, @AnthropicAI's Fable and Opus 4.8.
1
2
66
Francesco D’Orazio retweeted
IS THE ECONOMIST ALWAYS WRONG?
Scandalously, in some circles @TheEconomist has a reputation as a contrarian indicator. This week we fessed up to getting a big call on oil prices from April wrong.
Obviously our goal is not perfectly-hedged (and perfectly boring) predictive accuracy: often it is to stimulate, provoke, and challenge. But I did want to test that wider allegation, so I ran a series of LLM scorers across our full leader database since 2000 (7,000 leaders in all.)
You can see the results in the chart below: each dot is one of the 1,400 leaders where we identified concrete and falsifiable predictions that were central to the argument. Higher = more accurate, further to the right = more contrarian.
We do well, unsurprisingly, when aligned with conventional wisdom. We often do worse when truly out on a limb. But actually, on average, we are a bit likelier to be right than wrong on our somewhat-out-of-consensus calls. All round, a respectable performance.
And as @ecurrnomics points out an accompanying leader, there is no shame at all in being beaten by the market: as good free-marketers we believe deeply in the aggregated wisdom of prices.
Take a look at my piece here, which includes a canter through our best and worst calls of the last quarter-century: economist.com/interactive/fi…
78
218
1,570
680,116
Francesco D’Orazio retweeted
“Outdated business built a bot, A-I-A-I-O(bsolete).” 🐄🐖🐓🚜🤖
56
222
2,629
338,975
Francesco D’Orazio retweeted
Anthropic onboarding day: Michael Scott introducing Karpathy like he just signed Wemby in free agency.
394
1,470
17,655
2,474,470
We've shipped a major upgrade to our predictive crisis agent.
The new release gives Comms, PR and Brand teams something better than a flooded inbox of undifferentiated alerts:
→ A universal, dynamic four-stage threat framework: Signal, Narrative, Escalation, Crisis so you know exactly where a story sits on the escalation path.
→ Multi-dimensional risk assessment: Spread (who's talking), Intensity (how aggressively it's moving), Amplification (how far it's reaching).
→ Asymmetric Risk detection to flag coordinated activity early, without crying wolf at every spike.
→ A completely redesigned alert: human-readable cluster titles, the metrics that matter at the top, and richer evidence across channels.
Validated against thousands of narrative clusters across hundreds of brands. Live now for all Pulsar clients, no action required.
Full breakdown → lnkd.in/eBgB6QbW
Stop guessing. Start knowing. Introducing the new multi-dimensional Crisis Oracle.
Today, we’re excited to share Pulsar’s Upgraded Crisis Oracle, with sharper detection, threat staging and multi-dimensional risk scoring, informed by how teams have been using it. Detect, measure, and prioritize emerging threats with greater confidence.
audiences.pulsarplatform.com…
2
2
77
Francesco D’Orazio retweeted
Italy is the clearest example of how a rich country can slowly destroy itself through decades of bad management.
Since the late 1990s, Italy has had almost zero productivity growth.
Real wages have stagnated. And public debt is now around 137% of GDP, projected to move toward 139% overtaking Greece as the highest in the eurozone.
It is the result of 25 years of
1. Low innovation and very low adoption of new technologies
2. Family and insider succession in many companies, weakening meritocracy
3. Underdeveloped VC industry, startup ecosystem and R&D base
4. Political capture by lobbies and organised crime
5. Extreme administrative, judicial and business bureaucracy
6. Political fragmentation
7. Demographic collapse
No country can remain rich forever if productivity does not grow, debt keeps rising and politics only manages decline instead of reversing it.
108
338
1,230
80,857
American consumers when they get scammed: well played sir 🤓
Anthropic just ran the SaaS pricing playbook so cleanly that most people won't notice until their credit card statement arrives.
Step 1: Double usage limits for free over Christmas (Dec 25-31). Step 2: Double them again for two weeks in March as a "thank you." Step 3: Recalibrate everyone's baseline expectations upward.
Step 4: Today's announcement. Faster session burn during peak hours. The weekly cap stays the same on paper, but your 5-hour windows now drain quicker when you actually need them.
I spent $45 on Extra Usage today alone. Not because I wanted to. Because I was mid-workflow on a newsletter and the alternative was stopping for hours. That's the conversion event Anthropic is engineering.
The December doubling was framed as a holiday gift. The March doubling was framed as gratitude. Both were market research. They proved the infrastructure could handle 2x load during off-peak, then used that data to justify tighter peak-hour rationing. The 2x promo was a load test with good PR.
Thariq says ~7% of users will hit limits they wouldn't have before. That 7% is the highest-willingness-to-pay cohort. They're the ones who will click "buy extra usage" instead of switching to ChatGPT. Revenue per user goes up without touching the $20 sticker price.
This is the same move every mature SaaS company runs. Keep the base price friendly so nobody churns. Introduce metered overages so your power users self-select into higher spend. Salesforce, AWS, Snowflake, Twilio. Every infrastructure company eventually discovers that usage-based pricing on top of a subscription captures 3-5x more revenue per account than flat tiers alone.
Anthropic just figured out their version. The $20/month Pro plan is the front door. Extra Usage is the real product.
2
70