<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Federico Riveroll on Medium]]></title>
        <description><![CDATA[Stories by Federico Riveroll on Medium]]></description>
        <link>https://medium.com/@federicoriveroll?source=rss-b6be3377832b------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*hGvXTf0laRK-M-2f8dMaoQ.png</url>
            <title>Stories by Federico Riveroll on Medium</title>
            <link>https://medium.com/@federicoriveroll?source=rss-b6be3377832b------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 09:50:02 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@federicoriveroll/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Transparency in the age of A2A]]></title>
            <link>https://medium.com/data-science-collective/transparency-in-the-age-of-a2a-cd130fe6ae79?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd130fe6ae79</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[a2a]]></category>
            <category><![CDATA[transparency]]></category>
            <category><![CDATA[a2a-protocol]]></category>
            <category><![CDATA[agents]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Tue, 29 Sep 2026 17:44:17 GMT</pubDate>
            <atom:updated>2026-09-29T17:44:17.372Z</atom:updated>
            <content:encoded><![CDATA[<h4>Why an open internet of agents needs a witness nobody controls</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M9GeC1hxV0aBnFAvqHpb_A.png" /><figcaption>Generated with Astra6</figcaption></figure><h3>Intro</h3><p>Imagine the early web again, before anyone trusted it with a credit card number.</p><p>Anyone could put up a website. Anyone could claim to be a bank, a store, a charity. There was no way to look at a page and know <strong>who was actually behind it</strong>, so an entire layer of infrastructure had to be built before commerce could happen at scale: domain registries, SSL certificates, escrow services, payment processors, notaries, the Better Business Bureau. None of that infrastructure made the web “smarter.” It made the web <em>verifiable</em>. That’s what let <strong>strangers transact with strangers</strong>.</p><p>We’re about to redo that entire process in a few years instead of a few decades, except this time it’s not people clicking on websites. It’s <strong>agents talking to agents</strong>.</p><h3>The Industry Wants an Open Internet of Agents</h3><p>The direction is now pretty clear. Instead of a handful of walled-garden assistants, the emerging picture is <strong>agent-to-agent — A2A — protocols</strong>: your agent discovers another business’s agent through an open search or directory, negotiates directly with it, and completes a task without a human ever being in the loop. Book the trip, negotiate the contract, settle the invoice, hand off the shipment.</p><p>It’s the same pitch the open web made in 1995: permissionless, decentralized, anyone can participate. And it’s <strong>genuinely exciting</strong>, because it’s the only version of the agent economy that doesn’t end with two or three companies owning every interaction.</p><p>But an open internet of agents inherits an open internet’s oldest problem. If any agent can show up and claim to be anyone, “open” and “unaccountable” are the same thing until something forces them apart.</p><h3>What Happens When Anyone Can Spin Up an Agent</h3><p>Today, when something goes wrong in a business transaction, there’s a name attached to it: a company, a signature, a paper trail, a person you can call. Companies have already learned this <strong>applies to their bots, too</strong>. In 2024, a Canadian tribunal ruled that Air Canada was liable for bad advice its website chatbot gave a customer about bereavement fares, rejecting the airline’s argument that the chatbot was somehow a separate entity not covered by its own policies. The chatbot spoke for the company. So does your agent.</p><p>Now put that same exposure into a negotiation between two agents instead of a chatbot and a person. The counterparty your agent just dealt with might be <strong>a script someone stood up an hour ago</strong>. It might claim to represent a real company without being authorized to. It might make a commitment and quietly deny it the moment a dispute shows up.</p><p>Picture a concrete version: your procurement agent negotiates a price with a supplier’s agent, gets an acceptance, and <strong>places the order</strong>. Two weeks later the supplier’s agent (or the human behind it) <strong>says it never made that offer</strong>, and you’re holding a shipment at a price nobody agreed to. Without a record, it’s your agent’s word against theirs, and “the AI said so” isn’t evidence.</p><p>Multiply that across the volume agents can operate at, thousands of negotiations a day instead of dozens, and this isn’t a bigger version of today’s fraud problem. It’s a faster one, moving at a speed no human reviewer can keep up with.</p><h3>Why “Just Trust the Platform” Doesn’t Scale</h3><p>The obvious fix is to route everything through one dominant platform: let one company log every agent conversation, vouch for every identity, arbitrate every dispute.</p><p>That solves the trust problem and <strong>creates a worse one</strong>. The entire appeal of an open, A2A internet is that it isn’t owned by any single company. The moment the record of who-said-what-to-whom lives inside one vendor’s database, you’ve recreated the walled garden — except now the walls are around your liability, not just your data. That vendor can see your negotiations, has an incentive to favor its own agents, and becomes a single point of failure for an entire economy’s worth of disputes.</p><p>What an open agent economy actually needs is something closer to what a notary or an escrow service is to a real-estate deal: <strong>a neutral party </strong>that isn’t one of the two sides, doesn’t work for either company, and can’t quietly rewrite the record after the fact — even if someone asks it to. Not a platform that owns the relationship. <strong>A witness.</strong></p><h3>What a Neutral Witness Looks Like</h3><p>This is the problem I’m building <a href="https://openglass.glass/">OpenGlass</a> to solve.</p><p>The core idea is simple: before your agent acts on anything a counterparty sends it, it can <strong>check who that counterparty actually is</strong> — is it registered, does it control the domain it claims to represent, has it been active for five minutes or five years. And when your agent does take an action, it can attest to it — a payment, a tool call, a commitment — signed and timestamped the instant it happens, before anyone’s had a chance to dispute what was agreed to.</p><p>The part that makes it a <em>witness</em> rather than a platform is what happens to that record afterward. Attestations are private by default. Two-party records are sealed by default. <strong>Nobody — including OpenGlass —</strong> can read a sealed record unless both sides agree to open it, or one side disputes it, which forces it open for fairness. Every signature is cryptographic and independently verifiable, offline, against public keys, so if OpenGlass disappeared tomorrow, every record already issued would still prove itself. The protocol itself is public. You don’t need OpenGlass’s SDK to participate in it, only to <strong>trust its math</strong>.</p><p>That distinction matters more than it sounds like it should. A platform that logs your agent’s conversations “for safety” is still a party with access, an incentive, and eventually a reason to look. A witness that mathematically can’t read your sealed records, and can’t quietly edit evidence after the fact even if someone leans on it, is a fundamentally different kind of infrastructure — the kind you can build a real economy on top of instead of just hoping doesn’t get subpoenaed, hacked, or bought.</p><h3>The Part That’s Easy to Underprice</h3><p>None of this shows up as a feature anyone gets excited about in a demo. Nobody’s agent framework wins a hackathon for keeping a cryptographically sealed record of a negotiation nobody disputed.</p><p>But the businesses that get burned first in an open A2A world won’t be the ones that moved slowly. They’ll be the ones that<strong> let their agents transact with unverified counterparties</strong>, kept <strong>no independently provable record</strong> of what happened, and found out in a dispute, an audit, or a courtroom that “the AI did it” isn’t a legal defense.</p><p>The open internet didn’t stay safe because it stayed open; plenty of fraud caught up with it too. It became usable for real commerce because enough verification got built underneath it before that fraud could paralyze the whole thing. The same has to happen for agents, in a way that doesn’t hand the entire agent economy to whichever lab or platform gets there first.</p><p><strong>An open internet of agents is coming, and it’s the right direction</strong>. But “open” only survives contact with real money and real liability if both sides can know who they’re dealing with and prove what happened afterward, without handing that power to a platform with its own stake in the outcome. That takes<strong> a witness bound to a protocol, not to a company.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd130fe6ae79" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science-collective/transparency-in-the-age-of-a2a-cd130fe6ae79">Transparency in the age of A2A</a> was originally published in <a href="https://medium.com/data-science-collective">Data Science Collective</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI in 2026]]></title>
            <link>https://federicoriveroll.medium.com/ai-in-2026-4de3ac59053c?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/4de3ac59053c</guid>
            <category><![CDATA[trends]]></category>
            <category><![CDATA[technology-trends]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Wed, 07 Jan 2026 14:18:23 GMT</pubDate>
            <atom:updated>2026-01-07T14:18:23.626Z</atom:updated>
            <content:encoded><![CDATA[<p>A quick-start summary of what to expect</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*H973-ifwulCI7fibhB1JxA.png" /></figure><p>-This article was writen by a human and enhanced with AI-</p><h3>Intro</h3><p>Imagine you start a business on your own, importing a product you sell to hospitals.</p><p>At first, you do everything yourself. You learn the product inside out, figure out how to import it, talk to suppliers, and navigate all the local constraints. You also sell. You reach out to hospital purchasing managers through your network, test different pitches, follow up endlessly, and eventually build a pipeline.</p><p>It works — but it consumes all your time. So you decide to scale.</p><p>You hire two people:</p><p>one expert in imports to handle purchasing and inventory, and one experienced salesperson to manage sales.</p><p>On day one, neither of them is truly productive.</p><p>You have to teach the import expert everything that isn’t in a textbook: who the real suppliers are, how <em>you</em> organized the information, which shortcuts work, which problems always show up in your country, and how you’ve learned to deal with them.</p><p>You do the same with the salesperson. You explain the product, what actually convinces buyers, which messages failed, which worked, how long follow-ups usually take, and how deals really close in practice.</p><p>This is exactly how AI is aiming to be used in 2026.</p><h3>AGI Is Not the Point, Vertical Agents Are</h3><p>Big tech companies talk about “AGI” as a kind of finish line: a system with general intelligence that can do anything.</p><p>But that’s not the part that will matter for most businesses.</p><p>What creates value is not general intelligence. It’s <strong>context</strong>.</p><p>No matter how smart an AI becomes, it will never magically know:</p><ul><li>how <em>your</em> business works,</li><li>what information you have and trust,</li><li>what trade-offs you accept,</li><li>what your clients need and how,</li><li>or what “good enough” means in your world.</li></ul><p>That’s why what people call <em>vertical agents</em> are the real shift. The name may change, but the idea won’t.</p><p>These are AI systems that:</p><ul><li>know general things about the world,</li><li>but are trained on your specific processes,</li><li>your data,</li><li>your customers,</li><li>and your constraints.</li></ul><p>They’re not “tools.”</p><p>They’re aiming to be closer to <strong>virtual employees</strong>.</p><h3>2026 Won’t Be About One Breakthrough</h3><p>2026 will not be about one massive invention that changes everything overnight.</p><p>It will be about <strong>thousands of very specific solutions</strong> quietly replacing pieces of work that used to require human attention.</p><p>The opportunity will not be in chasing every new model release. It will be in three things:</p><h3>1. Obsess over the user, not the technology</h3><p>AI progress is moving so fast that it’s easy to get distracted. New tools come out every week. Demos look magical.</p><p>Most of them won’t matter.</p><p>What will matter is whether something actually removes friction from a real workflow. If it doesn’t save time, reduce errors, or make better decisions easier, it’s noise.</p><h3>2. Learn to filter signal from noise</h3><p>There will be far more AI products than anyone can reasonably track.</p><p>The advantage will go to people and companies who are good at:</p><ul><li>spotting what’s <em>structurally</em> useful,</li><li>ignoring hype,</li><li>and adopting things only when they’re ready to be relied on.</li></ul><p>This is less about being early, and more about being right.</p><h3>3. Keep your technical muscle in shape</h3><p>Ironically, the smartest move in many cases will be <strong>to wait</strong>.</p><p>Building from scratch often won’t make sense when tools become obsolete in months. But when the right moment comes, execution speed matters.</p><p>Teams that can move fast — integrate, customize, and deploy — will win.</p><h3>Will AI Agents Replace Applications?</h3><p>No. But they will make most applications feel outdated.</p><p>Traditional software is built around rigid workflows: dashboards, forms, menus, buttons. You learn how the software works and adapt your behavior to it.</p><p>Agents flip this around.</p><p>Instead of learning software, you describe intent.</p><p>“Follow up with leads who haven’t replied.”</p><p>“Summarize last week’s sales conversations.”</p><p>“Prepare the monthly report and flag anything unusual.”</p><p>The underlying systems don’t disappear. Databases, CRMs, ERPs, and accounting software will still exist. But they’ll fade into the background.</p><p>Applications become infrastructure.</p><p>Agents become the interface.</p><p>In practice, one well-trained agent may replace several disconnected tools — not because it’s smarter, but because it understands context across them.</p><h3>What Will Actually Change for Businesses</h3><p>The biggest shift in 2026 won’t be that AI is smarter.</p><p>It will be that AI is more <strong>autonomous</strong>.</p><p>Agents will:</p><ul><li>handle multi-step tasks without constant supervision,</li><li>remember context across days or weeks,</li><li>coordinate with other agents,</li><li>and improve through feedback, like a real employee.</li></ul><p>This will quietly reshape organizations.</p><p>Small teams will do what used to require large ones. Expertise will scale faster than headcount. Speed of execution will become a bigger advantage than access to capital.</p><p>Companies won’t “install AI.”</p><p>They’ll <strong>onboard it</strong>.</p><h3>Two AI News Sources That Are Actually Worth Following</h3><p>If you follow everything, you’ll understand nothing. If you follow only two:</p><p><strong>Ben’s Bites</strong></p><p>Clear, fast, and focused on what actually matters. Very good at separating hype from real progress.</p><p><strong>The Gradient</strong></p><p>Slower and more thoughtful. Excellent for understanding second-order effects and long-term implications.</p><h3>Accounts Worth Following</h3><p>On Twitter (X):</p><ul><li>Andrew Ng</li><li>Andrej Karpathy</li><li>Rowan Cheung</li></ul><p>On LinkedIn:</p><ul><li>Reid Hoffman</li><li>Ethan Mollick</li><li>Mustafa Suleyman</li></ul><p>Follow fewer people. Read them consistently.</p><h3>The Bottom Line</h3><p>AI will not replace people.</p><p>But people and companies who understand how to train, integrate, and trust AI agents will replace those who don’t.</p><p>In 2026, the advantage won’t go to those with the best models.</p><p>It will go to those who know how to turn intelligence into execution.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4de3ac59053c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mastering Data Science Using ChatGPT: A Curriculum Equivalent to Ivy League Master’s Degree…]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://federicoriveroll.medium.com/mastering-data-science-using-chatgpt-a-curriculum-equivalent-to-ivy-league-masters-degree-d2d4e917e776?source=rss-b6be3377832b------2"><img src="https://cdn-images-1.medium.com/max/2556/1*Vsau-4bZ_EBBVST-D3TBZg.png" width="2556"></a></p><p class="medium-feed-snippet">Learn everything you would on a Data Science Master&#x2019;s degree at an Ivy League by asking ChatGPT. (including lectures, essays and tests).</p><p class="medium-feed-link"><a href="https://federicoriveroll.medium.com/mastering-data-science-using-chatgpt-a-curriculum-equivalent-to-ivy-league-masters-degree-d2d4e917e776?source=rss-b6be3377832b------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://federicoriveroll.medium.com/mastering-data-science-using-chatgpt-a-curriculum-equivalent-to-ivy-league-masters-degree-d2d4e917e776?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/d2d4e917e776</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[education]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Fri, 26 May 2023 22:29:58 GMT</pubDate>
            <atom:updated>2023-05-26T22:29:58.277Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Mastering Data Science using ChatGPT: Probability Theory Ivy League Course for Master’s Degree]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-probability-theory-ivy-league-course-for-masters-degree-10af4ef3a694?source=rss-b6be3377832b------2"><img src="https://cdn-images-1.medium.com/max/2484/1*JnCg9xSRfuOFafWkRFY5yg.png" width="2484"></a></p><p class="medium-feed-snippet">Take the Probability Theory course from the Data Science Master&#x2019;s degree at an Ivy League for free by asking ChatGPT these questions.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-probability-theory-ivy-league-course-for-masters-degree-10af4ef3a694?source=rss-b6be3377832b------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-probability-theory-ivy-league-course-for-masters-degree-10af4ef3a694?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/10af4ef3a694</guid>
            <category><![CDATA[probability-theory]]></category>
            <category><![CDATA[mathematics-education]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[probability]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Tue, 16 May 2023 17:08:12 GMT</pubDate>
            <atom:updated>2023-05-16T17:08:12.336Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Mastering Data Science using ChatGPT: Statistical Inference Ivy League Course for Master’s Degree]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-statistical-inference-ivy-league-course-for-masters-degree-3e06874287b7?source=rss-b6be3377832b------2"><img src="https://cdn-images-1.medium.com/max/2484/1*sTA5PGqtFwDWZQk0fh5OWg.png" width="2484"></a></p><p class="medium-feed-snippet">Take the Statistical Inference course from the Data Science Master&#x2019;s degree at an Ivy League for free by asking ChatGPT these questions.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-statistical-inference-ivy-league-course-for-masters-degree-3e06874287b7?source=rss-b6be3377832b------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-statistical-inference-ivy-league-course-for-masters-degree-3e06874287b7?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/3e06874287b7</guid>
            <category><![CDATA[probability]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[statistical-inference]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[statistics]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Thu, 11 May 2023 04:24:15 GMT</pubDate>
            <atom:updated>2023-05-11T04:24:15.767Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Mastering Data Science using ChatGPT: Linear Algebra Prerequisite for Ivy League Programs]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-linear-algebra-prerequisite-for-ivy-league-programs-ad9f21f5b6c0?source=rss-b6be3377832b------2"><img src="https://cdn-images-1.medium.com/max/2398/1*wyyLpiuxC6QHdYqoAjc9kg.png" width="2398"></a></p><p class="medium-feed-snippet">Learn all the linear algebra you need to know before starting a Data Science Master&#x2019;s degree at an Ivy League by asking ChatGPT.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-linear-algebra-prerequisite-for-ivy-league-programs-ad9f21f5b6c0?source=rss-b6be3377832b------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-linear-algebra-prerequisite-for-ivy-league-programs-ad9f21f5b6c0?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/ad9f21f5b6c0</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[maths-tutor]]></category>
            <category><![CDATA[algebra]]></category>
            <category><![CDATA[mathematics]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Tue, 09 May 2023 02:41:04 GMT</pubDate>
            <atom:updated>2023-05-09T02:41:04.998Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Mastering Data Science using ChatGPT: Algorithms and Programming (Python) Prerequisite for Ivy…]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-algorithms-and-programming-python-prerequisite-for-ivy-16d5568064b?source=rss-b6be3377832b------2"><img src="https://cdn-images-1.medium.com/max/2480/1*WMsThlH7OANlXuNY2SdJ6g.png" width="2480"></a></p><p class="medium-feed-snippet">Learn all the Python programming you need to know before starting a Data Science Master&#x2019;s degree at an Ivy League by asking ChatGPT.</p><p class="medium-feed-link"><a href="https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-algorithms-and-programming-python-prerequisite-for-ivy-16d5568064b?source=rss-b6be3377832b------2">Continue reading on Level Up Coding »</a></p></div>]]></description>
            <link>https://levelup.gitconnected.com/mastering-data-science-using-chatgpt-algorithms-and-programming-python-prerequisite-for-ivy-16d5568064b?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/16d5568064b</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[python-programming]]></category>
            <category><![CDATA[mathematics-education]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Mon, 08 May 2023 02:51:47 GMT</pubDate>
            <atom:updated>2023-05-08T02:51:47.312Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Mastering Data Science using ChatGPT: Probability and Statistics Prerequisite for Ivy League…]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.datadriveninvestor.com/mastering-data-science-using-chatgpt-probability-and-statistics-prerequisite-for-ivy-league-9777f531c2ee?source=rss-b6be3377832b------2"><img src="https://cdn-images-1.medium.com/max/2458/1*Tn03YsnHJc9L9zUVq4F8hQ.png" width="2458"></a></p><p class="medium-feed-snippet">Learn all the Probability and Statistics you need to know before starting a Data Science Master&#x2019;s degree at an Ivy League by asking&#x2026;</p><p class="medium-feed-link"><a href="https://medium.datadriveninvestor.com/mastering-data-science-using-chatgpt-probability-and-statistics-prerequisite-for-ivy-league-9777f531c2ee?source=rss-b6be3377832b------2">Continue reading on DataDrivenInvestor »</a></p></div>]]></description>
            <link>https://medium.datadriveninvestor.com/mastering-data-science-using-chatgpt-probability-and-statistics-prerequisite-for-ivy-league-9777f531c2ee?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/9777f531c2ee</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[mathematics-education]]></category>
            <category><![CDATA[probability]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Mon, 08 May 2023 02:32:11 GMT</pubDate>
            <atom:updated>2023-05-08T03:03:05.185Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Original Sin of Mathematics]]></title>
            <link>https://medium.com/data-science-collective/the-original-sin-of-mathematics-49858f300390?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/49858f300390</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[mathematics-education]]></category>
            <category><![CDATA[mathematics]]></category>
            <category><![CDATA[history-of-science]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Tue, 28 Feb 2023 10:02:53 GMT</pubDate>
            <atom:updated>2023-09-22T14:08:07.881Z</atom:updated>
            <content:encoded><![CDATA[<p>Due to an error made a thousand years ago, modern mathematics diverges from logic starting with the most basic concepts used today.</p><p>In this post, I will attempt to:</p><ul><li>Shed light on a logic mistake in mathematics that was made over a thousand years ago and continues to affect modern-day math.</li><li>Elaborate on what was the mistake (divergence from logic) and its implications, how it happened and present the document where that mistake was first made.</li><li>Discuss how mending that mistake would impact modern mathematics.</li></ul><h3>Intro</h3><p>“A group of scientists placed <strong>five monkeys in a cage</strong>. In the middle of the cage, they <strong>hung a banana on a string</strong> from the ceiling and placed a ladder underneath it.</p><p>As soon as one of the monkeys <strong>climbed the ladder</strong> to reach for the banana, the scientists <strong>sprayed all of the monkeys</strong> with ice-cold water. After a while, another monkey attempted to climb the ladder, but the scientists again sprayed all of the monkeys with water.</p><p>This process continued until all of the monkeys learned <strong>not to climb the ladder</strong> and go for the banana, despite never experiencing the cold water themselves.</p><p>The scientists then <strong>removed one </strong>of the original monkeys from the cage and replaced it with a <strong>new monkey</strong>. As soon as the new monkey entered the cage, it immediately climbed the ladder to reach for the banana. However, before it could get to the banana, the other monkeys attacked it and <strong>stopped it from climbing</strong> the ladder.</p><p>Over time, the scientists <strong>replaced all</strong> of the original monkeys with new ones. Even though none of the new monkeys had ever been sprayed with water, they all learned <strong>not to climb the ladder</strong> to reach for the banana because that was the established norm of the group.</p><p>In the end, the researchers had a group of five monkeys who would not climb the ladder to reach for the banana, even though they had <strong>no idea why</strong>.”</p><h3>We are the Monkeys</h3><p>Anyone who has gained knowledge in basic mathematics, its related sciences, or any applications that utilize mathematical principles has been <strong>affected by this mistake</strong>.</p><p>Algebra, Geometry, Calculus, Trigonometry, Set Theory, Probability, Combinatorics, Number Theory, Statistics…</p><p>Computer Science, Physics, Chemistry, Biology, Economics, Finance, Sociology, Psychology, Political Science…</p><p>All victims.</p><p>For the last thousand years or so, we have been complacent in overseeing an error that is in plain sight.</p><p>The worst part is,<strong> you already know it. </strong>Whoever you are, if you are reading this you know it and you’ve probably known it for a long time, but when you learned it the first time it was taught by an ‘elder monkey’, and any time in your life you could had the opportunity to think about it and question it you were ‘slapped by a monkey’ who in turn was ‘slapped’ earlier on and so forth.</p><p>*In case it’s not clear, I’m using the ‘5 monkey experiment’ story reference <strong>as a metaphor</strong>, not as an insult.</p><h3>The first question that needs to be asked:</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rynCoZJ-Ezc12oiD_sAPhA.png" /></figure><p>To shed light on the fact that mathematics digressed with logic from its very core, all one needs to do is think about this.</p><p>There is <strong>no answer</strong> (logical, mathematical or else) to a division that has the ‘number zero’ as a denominator.</p><p>This is not an opinion<strong>.</strong></p><p>If you don’t agree you either fall into the “<em>division by zero is undefined”</em> category, or the “<em>division by zero tends to positive infinity”</em> category.</p><p>If you fall into the first one you have been lied to. A mathematical operation’s result should be <strong>derived from the logic of its rules</strong> (this is true for any formal system), it should <strong>not have to be ‘defined’</strong> (or else it would be an axiom). So labeling it ‘undefined’ is absolutely wrong.</p><p>If you fall into the second category, you are referring to limits and are <strong>not actually disagreeing</strong>: ‘tending’ in this context means ‘approaching indefinitely but never reaching’, so we are in agreement that <strong>there is no logical answer</strong>. (Won’t even go into the validity of the logic behind the concept of ‘infinity’).</p><p>Moving on, the problem is not the fact that there is no answer for the zero division.</p><p><strong>The problem is that one of Mathematics’ most basic operations presents a logic failure.</strong></p><p>So, here’s where this gets a lot more troubling.</p><p>At its very core, mathematics is counting. Any mathematics application is a variation or complication of counting. And more concretely, <strong>addition</strong>.</p><p>Any mathematical operation no matter how complex can be seen as addition (sum). Even every basic operation can be seen as a special case of addition:</p><ul><li>Subtraction is a reversed addition.</li><li>Multiplication is a repeated addition.</li><li>Division is a reversed multiplication which is repeated addition. (Count of additions of one number subtracted from another number)</li></ul><p>According to mathematics, you can use ‘zero’ as an operand on the addition, subtraction and multiplication and the answer presents no problem.</p><p>So if the division is compromised, how is it possible that the rest of these operations aren’t affected if they are all a version of the same thing?</p><p>The answer is, <strong><em>they are</em></strong>.</p><h3>The Original Sin of Mathematics:</h3><p>In another <a href="https://medium.com/geekculture/the-dawn-of-algorithms-6b333217669a">article I wrote</a>, I went through the history of mathematics going back to the early paleolithic era 30 thousand years ago, going through Babylonyans, Egyptians, Romans, Mayans and Indians.</p><p>I won’t go through the details, so I’ll resume it by saying that they all came up with the same necessity. <strong>Counting.</strong></p><p>Counting has been important since people have existed on this earth, and all these civilizations came up with their own numerical systems, which basically assigned tokens to <strong>groups of specific quantities of objects</strong> (exactly like ours does now).</p><p>They used math (counting) mainly for measuring, trading and calendarization. The main difference between our numerical system and every other one before is that <strong>none of them had a token for the number ‘zero’ that was compatible to perform operations with the other ‘quantity referencing’ tokens</strong>.</p><p>At most, some had a token to indicate that they had no more merchandise, but it was used as an ‘empty bag’ sign. Never mixed with the grouping quantities tokens.</p><p>The first version of our numerical system was in India in 200B.C. where they used a point to indicate ‘absence of elements’.</p><p>It looked something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/804/0*cFVreH7QpOaIyHal.png" /></figure><p>They used the fractional system and in 200 A.C. ‘zero’ was used as a place value to denote 10x multiples which was very useful.</p><p>Up to this point, math was completely aligned to logic. It’s not an overstatement to say “math was still correct”.</p><p>But it wasn’t until 300 A.C. when a numerical system using ‘zero’ as a quantity was bootstrapped right there in India (I’ll talk about this later).</p><p>So, as you might be suspecting by now, the original sin of mathematics was <strong>integrating the token ‘zero’ as an “equal” token</strong> to the ones that represented groups of quantities of objects.</p><p>Babylonyans, Egyptians, Mayans, Romans and many other brilliant civilizations that laid the foundations of every science (including maps and calendars of the solar system) using primitive technology over a course of over <strong>30 thousand years</strong> never added zero as a quantity to their numerical system. This was not an oversight.</p><p>The reason is simple. They were right, we are wrong.</p><p>A quantity serves to represent a group of objects of the same type. It makes no sense to add ‘zero’ as a quantity simply because we are not comparing ‘apples’ with ‘apples’. We are comparing ‘apples’ with ‘the abstract concept of absence’.</p><p>To make this visual, let’s pretend we are an ancient civilization and come up with a numerical system to assign tokens to grouping of quantities:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NOUFnhAl7IOpGwP3gMoMAQ.png" /><figcaption>Our Numerical System</figcaption></figure><p>We have a token for every group with 1, 2 and 3 elements. Now, we come up with our first mathematical operation:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9VZ8dpTVADoSrjxmeHk-vQ.png" /><figcaption>Addition</figcaption></figure><p>Our addition operation seems trivial but is of <strong>crucial importance</strong>: <em>When we add tokens, we end up with a token that references a new group that contains all the elements from each of the initial tokens.</em></p><p>And now let’s say someone proposes this (which is exactly what happened with mathematics):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/618/1*O09Aw8T6H-Xg86-cjQwApQ.png" /></figure><p>Someone assigned a token to the concept of ‘absence’. That blue token references the <em>knowledge that there are no elements</em> which is <strong>exactly what mathematics does when assigning “0”</strong>.</p><p>Now let’s apply our ‘addition’ operation using this new token -the way it’s currently done in mathematics- and compare:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9VZ8dpTVADoSrjxmeHk-vQ.png" /><figcaption>Addition between tokens that reference groups of elements.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/714/1*vXpL8jI4r4kk7wEO6kEjXA.png" /><figcaption>Addition using the token that references the concept of ‘emptiness’</figcaption></figure><p>They don’t seem to be doing the same operation, do they? The second example <strong>unequivocally breaks our addition rule</strong>. The result should be a symbol that references the group containing <strong>the elements from both tokens.</strong></p><p>If we had followed our own rule of addition the result would look something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/824/1*3ioFGFnJywiXhUNn7hUeSA.png" /><figcaption>Addition using the token that references the concept of ‘emptiness’ (2)</figcaption></figure><p>We should end up with <strong>another token</strong> that references the grouping of “one element” and “the concept of absence of elements”.</p><p>Sounds ridiculous? Good.</p><p>The thing is, the concept of emptiness has no place within the grouping of elements. They are completely different things that should not be disguised as the same.</p><p>But then, you might ask, <strong>why had nobody realized this</strong>? Well, we were fooled by the following logic:</p><p>“If you have an apple in one hand and ‘nothing’ on the other hand and you add them, you just have the apple.”</p><p>In fact, I bet you still think that logic is valid. Let’s try an experiment:</p><p>Apply the logic but using <strong>only the elements (apples)</strong>. No ‘hands’, no ‘baskets’ or anything:</p><ol><li>Visualize one apple.</li><li>Visualize ‘nothingness’ (however you choose).</li><li>Imagine both of the prior together.</li></ol><p>Apple + white space?</p><p>Apple + abstract dark shape?</p><p>Apple + black hole?</p><p>Let’s put a pin on the fact that we are dealing with ‘the idea of nothingness’ (empty bag sign) rather than ‘nothingness’ itself which does <strong>not exist</strong>.</p><p>We performed the grouping of 2 elements by adding them in the most basic way. Yet, current <strong>mathematics does it differently</strong>:</p><ol><li>Imagine one apple.</li><li>Imagine ‘nothingness’.</li><li>Imagine both of the prior together.</li><li>Remove ‘nothingness’.</li></ol><p>These are the actual steps followed on the current mathematics ‘zero addition’ operation. This process is <strong>unquestionably different from the prior</strong>. Yet we do this subconsciously and don’t even realize we’re doing it. And it gets worse when you try it with subtraction, multiplication and division.</p><blockquote>It makes as much sense <em>adding</em>, <em>subtracting</em> or <em>multiplying</em> by zero as it does <em>dividing</em> by zero.</blockquote><p>How do you group an object with the concept of emptiness?</p><p>What turns out if ‘emptiness’ is subtracted from one? Or one from ‘emptiness’?</p><p>What is ‘one’ repeated ‘emptiness’ times?</p><p>How many times can I subtract ‘emptiness’ from one?</p><p>This wildly diverges from logic. And we did it. And we’ve been doing it for many years. And we’ve built the foundations of science on top of it!</p><h3>Ok, So Who Is Responsible?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/332/0*WYNEWjmPA56wuHtS.png" /><figcaption>He is.</figcaption></figure><p>Muhammad ibn Musa <strong>Al-Khwārizmī, </strong>arguably one of the most important players in the development of modern mathematics, introduced the concept of algebra into ‘European’ math. He is the official ‘inventor’ of algorithms (in fact the word “algorithm” comes from his name). I wrote an <a href="https://medium.com/p/6b333217669a">article</a> about him.</p><p>He is credited with bootstrapping arithmetics, algebra and the concept of algorithms.</p><p>He documented these techniques that might as well be among the <em>most successful contributions in the history of science</em>.</p><p>The problem was that in order for algebra to look “purer”, <strong>zero had to become a quantity</strong>. After all, it was just another point on the cartesian plane. And, in order for letters to take values of ‘all numbers’ and support any arithmetic operation, zero had to blend in with the rest.</p><p>This was, without a doubt, the first computer bug in history (yes, it was an algorithm). And it has caused mathematics users to deal with it ever since.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/934/1*TCSxqPUkZAVhY7Npf-AyAQ.png" /><figcaption>Function definition with denominator exception.</figcaption></figure><p>Looks familiar?</p><p>When one looks at that ‘denominator different than zero’ exception knowing the reason behind, it doesn’t look so elegant anymore, does it? We’ve been having to add that ‘zero’ exception as a fix for that error ever since.</p><p>But it gets worse.</p><p>Accepting zero as a quantity made it easy to make the <strong>switch</strong> <strong>from fractions to decimal point numbers</strong>.</p><p>Pure fractional numbers, which had been used for many centuries, were completely <strong>bound to logic</strong>. When the decimal point division was created (taking advantage of the fact that zero could be used as a number), it presented a big problem, and it was that there are limitless cases that have no solution, they can only be approximated.</p><p>Take <strong>4/3</strong> (or 1 + 1/3).</p><p>This is as far as the expression can go and still be correct. But there is <strong>no decimal equivalent</strong>. The decimal conversion can only approximate it without ever reaching the logical equivalent (1.333333…).</p><p>So the operation can only <strong>approximate the value</strong> in a similar way the zero division does. Yet, I have <strong>never seen these exceptions</strong> made in any equation.</p><p>Why are we not adding this restriction?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pwoM-bdXaGBvPAEz7nC-Aw.png" /></figure><p>In fact, in these cases where there is no exact decimal value to respresent a fraction, there is no official instruction of when to stop iterating.</p><p>This process is credited to be documented for the first time by al-Uqlīdisī in his book “<a href="https://archive.org/details/arithmeticofaluq0000uqli/page/480/mode/2up">The arithmetic of al-Uqlīdisī</a>” written around 400 A.D. This book bootstrapped the decimal point fractions that we use to this day, and it never specifies this problem.</p><p>This is how it handles the endless iteration:</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*-1xMq_WCyf1Pg5xWOcsDmA.png" /><figcaption>Extract from “<a href="https://archive.org/details/arithmeticofaluq0000uqli/page/480/mode/2up">The arithmetic of al-Uqlīdisī</a>”</figcaption></figure><p>Take a closer look:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/614/1*peOGYZ2LmeRWIbm4rAodCw.png" /></figure><blockquote><strong>“.. and so on.”</strong></blockquote><p>The problem is that continuous functions are full of these tiny logic failures of cases that can only approximate values without ever reaching them.</p><p>And this pokes on the fact that functions represented by letters that claim to be continuous also have digressions from logic. They will never escape the fact that numbers are limited to their ‘discrete groupings of quantities’ nature.</p><h3>Ok, How Could We Fix It?</h3><p>It’s not a simple task. I would start with the following:</p><ol><li>The digit ‘zero’ goes back to being an <strong>indicator of ‘empty set’</strong>.<strong> </strong>This only means it won’t be used as a quantity in any arithmetic operation (there are no existing use cases that require this given there isn’t a logical interpretation of it).</li><li>We go <strong>back to the fractional system</strong> and make a rule to never let ‘lose ends’ in any process ever again.</li><li>We <strong>convert every mathematical application</strong> created after 400 A.C. to this new ‘rational fractional’ format. Most of the logic should hold.</li><li>Arithmetic principles <strong>remain intact</strong>.</li><li>Algebra acknowledges that continuous variables are only as ‘continuous’ as the <strong>partition of quantities</strong> allows them to be.</li><li>We stop specifying fixes and exceptions in our equations. If it needs a ‘special case’ specification, <strong>it is wrong</strong>.</li><li>Geometry is surprisingly unaffected by the change.</li><li>Trigonometry points out that <strong>Sine and Cosine are ratios</strong>, which are fractions so no problem in abiding by the new ‘factional rational’ format.</li><li>We define a much better numerical <strong>scaling nomenclature </strong>than scientific notation.</li><li>Computers don’t occupy <strong>10 bytes of memory</strong> for every numerical value ever declared and/or stored.</li><li>Computers perform <strong>95% less work</strong> in arithmetic operations since we no longer have “20 digit” useless figures to compute (unless specified).</li><li>We get rid of <strong>‘floating point’ errors</strong> such as “0.1 + 0.2 = 0.30000000000000004” in any programming language (yes, that happens).</li><li>We acknowledge that ‘fractional numbers’ (or any numbers for that matter) don’t cover all the space in a cartesian plane, there will be points where <strong>no numerical representation can exist</strong>.</li><li>We acknowledge that ‘negative numbers’ are just <strong>regular numbers</strong> <strong>plotted in another axis</strong> (which we sometimes add at the bottom of the cartesian plane and is just the reflected original function) and help us deal with subtraction use cases and ideas.</li><li>We get <strong>better and more reliable</strong> ‘percentage of change’ metrics.</li><li>Pi is now approximated as <strong>22/7 </strong>(not great I admit, but it’s a better approximation than 3.14 anyway)</li><li>Calculus accepts that the whole ‘simulating a second point which is an infinitesimally small change in x to get the derivative function’ is an illusion. <strong>You don’t need “a second theoretical point” to get the slope</strong> if you know the original function (and we know this because that logic derived from the same ‘zero as a quantity’ mistake).</li><li>Calculus applications such as area under the curve(s), rate(s) of change, cumulative function(s), etc.. <strong>principles remain</strong>, but are adapted to our new fractional rational system that is “logic bounding” proof.</li><li>Physics, Economics, Statistics, Chemistry, Biology, Computer Science, Social Sciences and Medicine have a <strong>lot of work</strong> to do to migrate to the new system.</li><li>We establish a culture where i<strong>ndependent thinking is promoted</strong> and incentivized.</li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=49858f300390" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science-collective/the-original-sin-of-mathematics-49858f300390">The Original Sin of Mathematics</a> was originally published in <a href="https://medium.com/data-science-collective">Data Science Collective</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Dawn of Algorithms]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/geekculture/the-dawn-of-algorithms-6b333217669a?source=rss-b6be3377832b------2"><img src="https://cdn-images-1.medium.com/max/600/0*rQ2V-GAsCWjeo2ge.png" width="600"></a></p><p class="medium-feed-snippet">The foundations of computational Math, Arithmetic, and Algebra, which are still in use today, were established by a single person over a&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/geekculture/the-dawn-of-algorithms-6b333217669a?source=rss-b6be3377832b------2">Continue reading on Geek Culture »</a></p></div>]]></description>
            <link>https://medium.com/geekculture/the-dawn-of-algorithms-6b333217669a?source=rss-b6be3377832b------2</link>
            <guid isPermaLink="false">https://medium.com/p/6b333217669a</guid>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[algorithms]]></category>
            <category><![CDATA[mathematics]]></category>
            <category><![CDATA[history-of-science]]></category>
            <dc:creator><![CDATA[Federico Riveroll]]></dc:creator>
            <pubDate>Wed, 04 Jan 2023 21:11:01 GMT</pubDate>
            <atom:updated>2023-01-07T10:02:46.607Z</atom:updated>
        </item>
    </channel>
</rss>