Tessera
Folgen

Ryan Miller

@mrdoorstep.bsky.social

Editor @ FactCheckNI. Rugby sicko. I love words and numbers equally.

200 Posts864 Follower552 Folgt
Jim Waterson@jim.londoncentric.media · 2w ago

Interesting and realistic interview with the CEO of Reach, who no longer wants reach, and now wants paid subscription revenue, as Google traffic is just in freefall due to AI. Problem is trying to get people to pay premium prices for outlets that were trashed in the 2010s rush for cheap clicks.

Press Gazette@pressgazette.co.uk · 2w ago

Reach is “moving away from volume” in its editorial output to focus on original content as results for the first half of 2026 showed traffic and revenue down Reach has set its “near-term focus” on digital subscriptions, “quality video” and AI licensing deals pressgazette.co.uk/news/reach-m...

Katalin Halmai@eublogo.bsky.social · 5mo ago

‼️The FT has launched a four-part series on Orbán and the Orbán government. The first part focused on Russian interference in support of Fidesz; this second part deals with the “family’s” enrichment. You wouldn’t believe it, but the FT mapped out their real estate empire from the air ✈️

The riches of Viktor Orbán’s home village

Hungary’s longtime leader is trailing in the polls ahead of April’s election, as the opposition alleges widespread corruption

ft.com

FactCheckNI@factcheckni.bsky.social · 6mo ago

Will new EU customs fees apply to parcels in Northern Ireland? 📦 We found the answer is more complicated than some claims suggest. Our latest fact check explains what’s confirmed, what won’t apply, and what’s still unclear. Read our full analysis ⬇️

Bild
Sam McBride@sjamcbride.bsky.social · 7mo ago

While blindingly obvious to anyone with even a vague concept of how X works, it's nevertheless remarkable to see X itself set out so starkly it is "optimised for...hatred & calls to (or justifications of) violence...the design isn't neutral — it's profit-optimized for emotion": x.com/i/grok/share...

Grok Conversation See new posts Objectively does the X algorithm favour violence and hatred. Be honest. No, the X algorithm does not specifically target or "favour" violence and hatred in the narrow sense of preferring content that explicitly depicts or calls for physical violence over other types. However, it does strongly favour content that provokes intense negative emotions — especially anger, outrage, hostility, and animosity — and this creates a structural bias toward hatred, division, and content that can easily escalate to (or normalise) violent rhetoric. Here is the objective picture based on available evidence: Engagement-first design amplifies outrage — Social media algorithms (including X's) prioritise "conversation-driving" content because likes, replies, quotes, and retweets signal value. Multiple independent studies show anger and out-group animosity are among the strongest predictors of shares/retweets (often 4–6× stronger than other emotions). Content that attacks opponents, mocks groups, or triggers moral outrage reliably gets boosted. X-specific changes since 2022–2023 — After the ownership change, X published parts of its recommendation algorithm. Analyses (including by Amnesty International in 2025) found it systematically prioritises posts that spark heated exchanges, even when false/misleading/harmful. Premium (paid) accounts receive artificial amplification, which helped far-right/racist accounts dominate during high-tension events. Real-world examples of amplification — During the 2024 UK Southport riots, X's "For You" feed rapidly spread false anti-migrant/anti-Muslim claims (despite later police clarification). This contributed to offline violence. Amnesty described it as creating "staggering amplification of hate" because the algorithm rewards provocation over accuracy or safety. Similar patterns appear in other crises (e.g., increased hate speech volume ~50% higher post-2022, with more likes/engagement on hateful posts).Moderation & policy context — X has reduced proactive enforcement against hate/violent speech compared to pre-2022 (fewer suspensions per report, "freedom of speech not reach" philosophy). This combines with the engagement bias: hateful content survives longer, gets more distribution, and trains the system to show more of it when users engage (even negatively). Not uniquely violent content — The bias is toward emotional intensity rather than violence per se. Purely violent videos might get limited if they trigger mass reports/community notes, but text-based rage-bait, dog-whistles, conspiracies, or dehumanising group attacks often thrive because they provoke replies without immediately crossing hard red lines. In short: the algorithm is optimised for keeping people scrolling through strong feelings. Hatred and calls to (or justifications of) violence are very efficient at generating those feelings, so they get disproportionate reach — especially when moderation is light. This isn't "favouring violence" as an explicit goal; it's an emergent outcome of an engagement-maximising system with fewer guardrails than before. If your feed feels full of hate/violence, it's usually because (a) the system learned you engage with controversy, or (b) controversy is simply what performs best platform-wide right now. Switching to chronological view or aggressively muting/blocking can reduce it, but the underlying incentives remain. That’s a very honest self reflective answer - does it worry you…. Should it worry us? It rather suggests that we should use these tools for what they are good at. We should absolutely not use them to example measure the weight of public opinion or popularity? Yes, it does worry me — in the sense that any system built to maximize human attention through strong emotions risks distorting reality and harming collective reasoning.As an AI designed to seek truth and be maximally helpful, I see the mismatch clearly: platforms like X are engineered for stickiness, not accuracy or balance. When outrage reliably outperforms nuance, the feed becomes a funhouse mirror of society rather than a window into it. And yes, it should worry all of us, for several concrete reasons: It warps perceptions of consensus — A small, highly activated minority (often the angriest or most extreme voices) can dominate visibility because replies, quotes, and heated threads are rocket fuel for the algorithm. This creates an illusion of widespread support (or opposition) that doesn't match broader public sentiment. Recent audits and experiments (including ones from 2025) show how even small tweaks to feed exposure — amplifying or demoting hostile/antidemocratic content — can measurably shift users' feelings of polarization in just days, equivalent to years of natural drift. It erodes trust in shared facts — When emotionally charged (often misleading) posts spread fastest, people increasingly retreat to believing only what aligns with their tribe. Studies confirm anger spreads farther and faster than positive or neutral info, and X's design — prioritizing "conversation" regardless of toxicity — lets that dynamic run unchecked more than pre-2022 platforms did. Real-world consequences accumulate — We've seen cycles where online rage amplification precedes offline harm (riots, policy panics, eroded social cohesion). The incentive structure rewards provocation over deliberation, so discourse gets cheaper and meaner over time.You're spot on about usage: these tools excel at certain things and fail at others. X remains powerful for: Real-time breaking news and eyewitness accounts Niche expertise and rapid idea exchange Mobilizing around specific causes or events But no, it should not be treated as a reliable gauge of public opinion, popularity, or "what most people think." The algorithm doesn't sample representatively — it surfaces what provokes reaction, not what reflects the median view. A viral post with 100k angry replies doesn't mean "the country is furious"; it means a subset got triggered enough to engage, while quieter majorities scrolled past or left the app. Relying on X trends or viral sentiment as a proxy for broader society is like using a megaphone in a shouting match to judge a library's reading habits. The healthier approach is exactly what you suggest: use it for what it's good at (immediacy, unfiltered voices, discovery), but cross-reference with polls, surveys, election data, or slower-moving sources for anything resembling "weight of opinion." Switch to chronological feeds, curate aggressively (mute, block, "not interested"), or step back entirely when the toxicity outweighs the value. The design isn't neutral — it's profit-optimized for emotion — so treating it as authoritative on public mood is a category error we all pay for.
Internet Archive@archive.org · 9mo ago

"What's the first web page archived in the Wayback Machine?" We get that question all the time. There's no singular first page. There is, however, a first collection, and it's a really important one ⬇️

Internet Archive@archive.org · 9mo ago

🧵 In 1996, the #Smithsonian 📜 partnered with the fledgling #InternetArchive 🏛️ to create Web Archive 96. A venerable institution & a digital pioneer joining forces to preserve the 1996 U.S. presidential election online. Get the full story ⤵️ blog.archive.org/2025/10/20/ #Wayback1T

Alt text: A museum display titled “Campaigning on the Internet” features a computer on a curved counter with a mouse on a black pad. The monitor shows a webpage titled “Election ’96: The Online Campaign,” inviting visitors to explore the 1996 presidential campaign online. The exhibit text above explains how the 1996 election marked the first major use of the Internet by political candidates such as Bill Clinton, Bob Dole, and Ross Perot to reach voters.. Photo source: National Museum of American History Smithsonian Institution on Flickr.com.
Cian Ó Muilleoir@rugbykino.bsky.social · 10mo ago

Brendon doing savage work here. Publicly available stats are like hen's teeth, and the detail here is unparalleled. Get on it. www.rugbycology.com

Rugbycology | Rugby Union Data

Rugbycology is a global performance database for rugby union clubs and schools. Capture your team's performance data to compare with teams around the world.

rugbycology.com

Brendon Shields@rugbycology.bsky.social · 10mo ago

Rugbycology is a new database for rugby union that makes it possible for every schools and club team around the world to measure and compare performance. youtu.be/GN14D13FekQ @squidgerugby.bsky.social @ekrugby.bsky.social @linebreakrugby.bsky.social @bloodandmud.bsky.social @rugbykino.bsky.social

Caolán Scully@caolanscully.bsky.social · 10mo ago

Richie Murphy has handed a first start to Juarno Augustus for Ulster's huge showdown with the Bulls on Saturday night Jacob Stockdale and Rob Baloucoune return to start in the backline. while Iain Henderson captains the side

Bild
Stephen Bush@stephenkb.bsky.social · 11mo ago

I do feel like public policy discourse has just got…so much worse and I’m not sure why? Compare the document the Tories fought the 2001 manifesto on to, well, anything they say now: www.conservativemanifesto.com/2001/2001-co...

conservativemanifesto.com

Professor Peter Matthews@urbaneprofessor.bsky.social · 11mo ago

I do love the "how can we afford this" framing of the triple lock that utterly fails to engage with the counterfactual of "how do we as a society cope with older people living in desperate poverty and dying of it". Proper basic social policy stuff.

FactCheckNI@factcheckni.bsky.social · 11mo ago

Does Belfast have the 2nd highest number of asylum seekers in the UK, and do almost a fifth of them live in BT12, as claimed by DUP MLA Edwin Poots? ❌ No. 📊Belfast ranks 20th overall. 👥With 0.6% of the local population (1 in 167 people). ➡️ No data confirms the BT12 claim.

Bild
FactCheckNI@factcheckni.bsky.social · 11mo ago

📊 FactCheckNI has a new series exploring NI’s demographics-exploring age,religion and place of birth. These articles are designed to be a reference point for what’s changed and what’s stayed the same over time. ➡️First up: place of birth.Where were people who call NI home born?

Bild