2c:null 2d:["$","title",null,{"children":"Charting the Great Twitter Bot Purge of 2018 | Jeremy Ashkenas | Observable"}] 32:Ta96,md`# Charting the Great Twitter Bot Purge of 2018 Today, Twitter [began purging its platform of tens of millions of fake accounts](https://www.nytimes.com/2018/07/11/technology/twitter-fake-followers.html). In January, The New York Times [published an investigation](https://www.nytimes.com/interactive/2018/01/27/technology/social-media-bots.html) into the practice of buying fake Twitter accounts, exploring the phenomenon through the lens of Devumi — a single company that specialized in selling realistic-looking followers — bots masquerading behind the faces of real people, whose names and photographs had been scraped from Twitter. Celebrities like [Kathy Ireland](https://twitter.com/@kathyireland) and [Paul Hollywood](https://twitter.com/@paulhollywood), businessmen like [Michael Dell](https://twitter.com/@michaeldell), actors like [Ryan Hurst](https://twitter.com/@rambodonkeykong) and [John Leguizamo](https://twitter.com/@johnleguizamo), the New York Times’ "Frugal Traveler" [Lucas Peterson](https://twitter.com/@lucaspeterson), and even [Martha Lane Fox](https://twitter.com/@marthalanefox), a prominent member of Twitter's Board of Directors, were all caught with thousands of purchased fake followers. In response, [a criminal investigation was opened](https://www.nytimes.com/2018/01/27/technology/schneiderman-social-media-bots.html) by the attorney general in New York, and [Twitter quickly deleted the Devumi bots](https://www.nytimes.com/interactive/2018/01/31/technology/social-media-bots-investigations.html). [Unilever declared war on fake accounts.](https://www.mercurynews.com/2018/02/12/facebook-google-twitter-could-lose-unilever-ads/) Now, six months later, [a much wider purge has begun](https://blog.twitter.com/official/en_us/topics/company/2018/Confidence-in-Follower-Counts.html). Here are ** ${ Object.keys(rawdata).length } charts showing follower counts over the past year** for a mix of Twitter accounts, some which had been caught with fake followers by the Times investigation in January, along with a dozen other celebrities. Kathy Ireland lost nearly 80% of her online following today, and Michael Dell and Martha Lane Fox are down about halfway from their peaks. Jack Dorsey lost more followers than he's gained over the past year. While Elon Musk and Cardi B have mostly escaped thus far... To play fair, I threw myself in there too. If you notice other interesting Twitter accounts who have lost large numbers of followers today, tweet them to me [\`@jashkenas\`](https://twitter.com/jashkenas), and I'll add them to these charts tomorrow. *July 12, 2018
Source: [Trendsmap](https://www.trendsmap.com/) *

`33:Ta46,function minichart(handle) { const zeroed = show === "Scaled for all followers"; const data = rawdata[handle].map(d => d[1]); let w = width > 800 ? width / 4 : width > 600 ? width / 3 : width > 450 ? width / 2 : width; let s = d3.select(svg``); let margin = { top: 5, right: 80, bottom: 5, left: 0 }; w = w - margin.left - margin.right; let h = +s.attr("height") - margin.top - margin.bottom; let g = s .append("g") .attr("transform", "translate(" + margin.left + "," + margin.top + ")"); var x = d3.scaleLinear().rangeRound([0, w]); var y = d3.scaleLinear().rangeRound([h, 0]); if (zeroed) { var area = d3 .area() .x((d, i) => x(i)) .y1(d => y(d)) .y0(h); var line = area.lineY1(); } else { var line = d3 .line() .x((d, i) => x(i)) .y(d => y(d)); } var yExtent = d3.extent(data); x.domain([0, data.length]); y.domain(zeroed ? [0, yExtent[1]] : yExtent); const lastPoint = data[data.length - 1]; const peakLoss = lastPoint - yExtent[1]; const peakLossPct = peakLoss / yExtent[1]; const yearLossPct = peakLoss / (yExtent[1] - data[0]); const ticks = [zeroed ? 0 : yExtent[0], lastPoint]; if ((zeroed ? peakLossPct : yearLossPct) < -0.1) ticks.push(yExtent[1]); g.append("g") .attr("transform", `translate(${w + 5}, 0)`) .call( d3 .axisRight(y) .tickValues(ticks) .tickFormat(d => (d ? d3.format(".2s")(d) : 0)) ) .select(".domain") .remove(); if (zeroed) { g.append("path") .datum(data) .attr("fill", "steelblue") .attr("fill-opacity", 0.2) .attr("d", area); } g.append("path") .datum(data) .attr("fill", "none") .attr("stroke", "steelblue") .attr("stroke-linejoin", "round") .attr("stroke-linecap", "round") .attr("stroke-width", 2) .attr("d", line); g.append("circle") .attr("cx", x(data.length)) .attr("cy", y(lastPoint)) .attr("r", 3) .attr("fill", "steelblue"); const absolute = d3.format(".2s"); const pct = d3.format(".0%"); const div = html`

${s.node()}

@${handle}
${ peakLoss < 0 ? `${absolute(peakLoss)} followers
Lost ${pct(-peakLossPct)}` : `No loss` }

`; return { html: div, loss: peakLossPct }; }34:[] 30:{"id":"a0eceadc1bc4a274","slug":"charting-the-great-twitter-purge-of-2018","title":"Charting the Great Twitter Bot Purge of 2018","thumbnail":"ed276287f22553844f4b3f4279575b0e60e44285bfa500d67c11c643c58d76c2","access_level":"public","create_time":"$D2018-07-12T22:40:05.812Z","update_time":"$D2018-07-13T23:35:56.265Z","view_time":null,"first_public_time":"$D2018-07-13T01:23:16.296Z","has_unpublished_edits":null,"fork":false,"stars":15,"starred":false,"owner":{"type":"team","id":"172cc08ef8a18a4b","login":"jashkenas","name":"Jeremy Ashkenas","avatar_url":"https://avatars-next.observableusercontent.com/fa1903f8171d4621ef49de28e0ff1fb965960ec87df175d8b543c4b09508b45b","tier":"starter_2024","bio":"Used to work on @observablehq — still think it rocks.\n🏍 🌎 👶"},"version":347,"publish_version":347,"role":null,"authored":false,"authors":[{"type":"individual","id":"783616ba9fb55f5d","login":"jashkenas","name":"Jeremy Ashkenas","avatar_url":"https://avatars-next.observableusercontent.com/fa1903f8171d4621ef49de28e0ff1fb965960ec87df175d8b543c4b09508b45b","tier":null,"bio":"Used to work on @observablehq — still think it rocks.\n🏍 🌎 👶"}],"members":[{"type":"individual","id":"783616ba9fb55f5d","login":"jashkenas","name":"Jeremy Ashkenas","avatar_url":"https://avatars-next.observableusercontent.com/fa1903f8171d4621ef49de28e0ff1fb965960ec87df175d8b543c4b09508b45b","tier":null,"role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":5,"collection":null,"type":"notebook","body":{"titling":"auto","title":"Charting the Great Twitter Bot Purge of 2018","cells":[{"id":0,"value":"$32","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":325,"value":"viewof show = radio({\n options: [\"Scaled to past year\", \"Scaled for all followers\"],\n value: \"Scaled to past year\"\n})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":27,"value":"html`

${Object.keys(rawdata)\n .map(handle => minichart(handle))\n .sort((a, b) => a.loss - b.loss)\n .map(c => c.html)}

`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":20,"value":"$33","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":322,"value":"import {radio} from \"@jashkenas/inputs\"","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":21,"value":"d3 = require(\"d3\")","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":17,"value":"rawdata = (await fetch(\n \"https://gist.githubusercontent.com/jashkenas/17aad4d081412e1083156d2861eee55e/raw/b1b4b5051ffcb038602fb2601af33978a0970671/followers.json\"\n)).json()","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W34","stdlib":"1"}} 31:"$30:body"

@${handle} ${ peakLoss < 0 ? `${absolute(peakLoss)} followers Lost ${pct(-peakLossPct)}` : `No loss` }

@${handle}
${ peakLoss < 0 ? `${absolute(peakLoss)} followers
Lost ${pct(-peakLossPct)}` : `No loss` }