<?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 Joshua McKenty on Medium]]></title>
        <description><![CDATA[Stories by Joshua McKenty on Medium]]></description>
        <link>https://medium.com/@jmckenty?source=rss-aa6edb551662------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*B6elyDM3KdmXS4jBtics5w.png</url>
            <title>Stories by Joshua McKenty on Medium</title>
            <link>https://medium.com/@jmckenty?source=rss-aa6edb551662------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 10:29:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@jmckenty/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[A Proposed Policy Framework for Synthetic Media]]></title>
            <link>https://medium.com/@jmckenty/a-proposed-policy-framework-for-synthetic-media-532d2d4b7791?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/532d2d4b7791</guid>
            <category><![CDATA[free-speech]]></category>
            <category><![CDATA[policy]]></category>
            <category><![CDATA[medium]]></category>
            <category><![CDATA[deepfakes]]></category>
            <category><![CDATA[regulation]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Fri, 15 Dec 2023 20:58:41 GMT</pubDate>
            <atom:updated>2023-12-15T20:58:41.184Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="https://www.deepactual.com/"><em>Joshua McKenty</em></a><em> — DeepActual<br></em><a href="https://www.linkedin.com/in/khadem-badiyan-18364912a/"><em>Khadem Badiyan</em></a><em> — DeepActual</em></p><p>None of the harms done (or possible) through synthetic media are unique to the methods of producing that media, and none of them can be prevented <em>strictly </em>through control of the technology of its production. Thus, a policy framework for addressing the risks of harmful synthetic media must be a policy framework for digital media writ large. While we have seen several policy proposals in recent weeks, most of them have been narrowly focused, with potentially huge unintended consequences. This draft is intended to highlight a few missing pieces, and to spark further conversation.</p><figure><img alt="Black and white clip from the movie Mr. Smith Goes to Washington, showing the senator holding letters in each hand in the Senate; He looks haggard and defeated." src="https://cdn-images-1.medium.com/max/550/0*Q5d1mONh0dbEZh7N" /><figcaption>Mr. Smith Goes to Washington</figcaption></figure><h3>The Audience for this Proposal</h3><p>The harms of synthetic media are global in both source and target, and the stakes could not be higher. But the legislative landscape, as well as the landscape of public opinion, is hardly uniform. We have endeavored to envision a policy that could be adapted to, and adopted by, national and regional governments in North America, Europe, and India at the very least. (Many such governments are passing relevant regulations as we speak.)</p><h3>Goals of this Framework</h3><p>The goals of this framework are not to eliminate the creation of deepfakes (or any other form of synthetic media). They are:</p><ul><li>To reduce the scope of harms, and…</li><li>To make authorship explicit, allowing us…</li><li>To push the underlying harms back into the jurisdiction of existing regulations…</li></ul><p><em>…and to achieve all of the above with the utmost SPEED.</em></p><p>To be specific, there are real harms both in suppressing media suspected of being synthetic when it is not, and in allowing synthetic media to masquerade as actual. This proposal strives, not only to balance those harms, but to defer final judgment of such balance to the judiciary function in the appropriate jurisdiction as swiftly as possible.</p><p>It is additionally our goal <strong>not </strong>to propose any policy which would challenge or shift current regulation around freedom of speech, political expression or foreign interference.</p><h3>What Media Would Be Covered Under This Policy?</h3><p>While it is conceivable that synthetic media will be put to every possible (good and evil) use, the specific harms that have already been observed are the following:</p><blockquote>Domestic Political Interference<br>Identity Theft &amp; Fraud<br>Non-consensual Pornography<br>Foreign Political Interference<br>Military Disinformation Campaigns<br>Insurance Fraud<br>False Evidence</blockquote><p>Any media (photographs, video, audio recordings, whether publicly broadcast or privately circulated) that attempts or achieves these harms, is within the scope of this policy.</p><h3>Framework for Media Policy</h3><p>This framework must be shaped around the complete lifecycle of media, from capture or creation through publication to reporting and mitigation, and address the various actors involved.</p><figure><img alt="Sticky notes showing seven policy areas: Technology, Technicians, Content, Distribution, Claims Handling, Mitigation, and Deterrents." src="https://cdn-images-1.medium.com/max/1024/1*4C7awWwGCr2b5NPOh_EFsQ.png" /><figcaption>A framework for media policy</figcaption></figure><h3>1. Technology of synthetic media creation</h3><p><strong>We <em>could</em> try and regulate the use of “strong” deepfake approaches</strong> in the same way we regulate strong encryption (as a munition, and covered by ITAR); however, it is likely too little, too late for this approach. Today’s deepfakes technologies are already the AR-15 of synthetic media generation, and there is no natural technology “line” (resolution, frame-rate, amount of computing power per pixel or frame, complexity of training data set, etc.) that has not already been crossed…</p><p>… the cat is already out of the bag.</p><p>Instead, we must <a href="https://medium.com/@jmckenty/the-case-for-verified-media-3f0f5bb83fd1"><strong>establish standards for media provenance</strong></a>, and require media that causes harm to defend its provenance. This mirrors policy that has been applied to the questions of <a href="https://en.wikipedia.org/wiki/WIPO_Copyright_Treaty">copyright under the DMCA/ISD</a> — once a claim of copyright infringement is made, the publisher of the media is obligated to provide evidence of their right to use the infringing media. When they have done so, the burden of proof shifts to the <strong>claimant</strong> to discredit those rights.</p><p>The <a href="https://medium.com/@jmckenty/the-case-for-modular-media-verification-bf93e327bb40">minimum provenance for media must establish <strong>when, where, and with what device</strong></a> the original source media was captured. Constructed media that is not an edit of some original source media does not require provenance, but must not claim to be anything but synthetic.</p><p><em>(We acknowledge that tooling for capture and authentication of media provenance is not yet widely available; in the short term, the simplest form of provenance is a sworn affidavit of the media producer.)</em></p><h3>2. Duties of Technicians</h3><p>While one of the biggest threats of today’s media landscape is the growing ease with which amateurs, using small budgets and open source or freely available tooling, can produce compelling and confusing synthetic media, crafting truly undetectable deepfakes is (and will likely remain) the domain of experts, at least for now.</p><p>Obviously, criminals and foreign agents have no interest in, or requirement for, licenses or certifications. But the majority of professionals who develop deepfakes for legitimate purposes (movies, advertising, comedy) could be expected to subscribe to a standard of business conduct, attached to an oversight body. (This would be similar to how tradespersons, doctors, lobbyists, accountants and other professionals are governed). Any involvement in harmful media would put their ability to participate in lucrative, legitimate synthetic media at risk.</p><p>There has been no successful attempt to create a supervisory body for software professionals yet — if such a licensing or certification scheme is attempted in the context of AI, it would make sense to include media synthesis in its scope. Required certifications should be aligned with the size and scope of the company involved, similar to the difference between a “handyman” and a Red Seal carpenter.</p><h3>3. Content Policies</h3><p>We have a single, basic policy recommendation at the content level: <strong>Any media that generates a synthetic representation of a real person or contemporary event, must be labeled as synthetic.</strong></p><p>Establishing such a labeling policy (and working closely with the major distribution platforms on the user experience of such labeling) will avoid most of the potential overreach of this policy framework, and preserve the rights to use synthetic media of all forms for satire and political commentary. The <a href="https://www.congress.gov/118/bills/hr5586/BILLS-118hr5586ih.pdf">DEEPFAKES Accountability Act</a> is a reasonable starting point for such labeling, and could be expanded with more depth on metadata storage and verification.</p><p>We believe such <strong>labeling must be within the media itself</strong>, and not just within the distribution platform. This will help to address the uneven regulatory landscape of distribution (what we have called the TikTok Challenge), as well as the major role that private, walled-garden communication groups now play in the circulation of misinformation (the WhatsApp Gap).</p><p>Of course, many of the harms of such content are not mitigated by simply labeling it — but we believe <em>all of those other harms have existing regulations</em> that address them. Beyond labeling, what is required is an <strong>expedited,</strong> <strong>low-friction mechanism </strong>for enabling and supporting those existing law enforcement efforts. This could include<a href="https://hbr.org/2021/08/its-time-to-update-section-230"> updates to Section 230</a> that would apply “duty of care” for unlawful content, such as has been suggested in Congressional testimony.</p><p>The second category of harmful media is media that has been substantially edited, such that the meaning and content of the original media is altered or obscured. Yet most media is edited before publication. In order to treat the content of “trivially edited” media as equal to the original source, we must agree on the definition of “substantially edited”, which might include changes to frame-rate, obscuring or inserting context, or selectively removing frames of video or audio. Substantially-edited media should have the same labeling requirements as synthetic media.</p><h3>4. Distribution of synthetic media</h3><p>It is likely impossible for any major media distribution platform to implement an objective, non-partisan content review process. (Certainly it is unlikely to happen under voluntary self-governance.) Even if it were possible, it will remain impossible to audit or confirm externally, and is therefore impossible to trust.</p><p>Fortunately, content review is not required. (As Linus Torvalds famously said, “With enough eyeballs, all bugs are shallow.”) What <strong>is</strong> required is a robust obligation to participate in extended reporting and mitigation programs (such as the existing <a href="https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/europe-fit-digital-age/digital-services-act_en#:~:text=Digital%20Services%20Act%20(DSA)%20overview,and%20the%20spread%20of%20disinformation.">UK Digital Services Act</a>, etc). These mitigation programs will require platforms to support verification of emerging media provenance standards, and of publisher identity (already a common feature).</p><p>Specific policies for reporting and mitigation must include deadlines in hours or days, not weeks or months.</p><h3>5. Handling Claims of Harm</h3><p>Much like the DCMA process, requests for mitigation of harmful media must be submitted by those with “standing”. This is most simply the individual or a member of the group harmed, or a legal proxy. Outside of harms where an existing government agency is actively involved (such as election interference or foreign disinformation campaigns), it is likely inappropriate for the government to play a major role in monitoring and reporting content.</p><p>When a claim of harm is submitted (either to law enforcement, or directly to a media distribution platform), the claim receiver must contact the media publisher with a “demand for provenance”. Depending on the severity of the harms being claimed, the bar for establishing provenance could vary from a publisher’s affidavit, to authenticated metadata or chain-of-custody. It is important that the bar is based only on the <strong>stakes</strong> of the claim, not of the claimant or the publisher.</p><p>As long as the media publisher can provide provenance meeting this initial bar, their media remains available and unaltered; the burden of proof is now on the claimant to demonstrate that their provenance is false or insufficient (and the bar to prove false provenance must be much higher than that presented to the publisher). If it happens that the claimant can meet such a bar (or if the media publisher fails to provide provenance in the first place), the harmful media must be mitigated by the platform.</p><p>The timelines for notice and response must be proportional to the rate that the media is being viewed; in cases where harmful synthetic media is trending across social media platforms and being viewed by millions of people hourly, provenance must be demonstrated or asserted within hours. And in cases where claims of harm are extremely likely (such as political advertisements, media depicting famous persons, political figures, events in war, or pornography), the platform might choose to require provenance for initial media publishing.</p><p>Media platforms should maintain resources available for addressing and handling claims in proportion to the rate of claims within that region, topic, or community.</p><h3>6. Mitigation of Harmful Media</h3><p>As is well documented elsewhere, <em>removal</em> of false media is often the wrong approach. In the case of public access to edited content such as a deepfake, platforms have an obligation to redirect viewers to the unedited original content, describing <em>but not displaying</em> the harmful edits. Synthetic media that is intended as commentary or satire should have the option of applying appropriate labeling as a simple remedy.</p><p>For harmful media that was constructed entirely without an original source, where the publisher is unwilling to comply with labeling or where the media is unlawful regardless of labeling, the media should be replaced with an alternative media provided by the claimant.</p><figure><img alt="Detour sign on a stretch of empty road." src="https://cdn-images-1.medium.com/max/1024/0*AvxCXxdu7XJEVJEo.jpg" /><figcaption>Mitigation should emphasize redirection, not removal</figcaption></figure><p>Media platforms often have methods to notify viewers of new or updated content; they should use these methods to notify viewers who were exposed to the original harmful media of its synthetic nature, and provide access to the redirected or replacement content if relevant. Similarly, private communication platforms such as WhatsApp, Instagram Group Chats, etc. should send an additional message to chat group members (or previous members) with notification.</p><p>We should treat any mitigation of harmful media as essentially the enforcement of a court injunction, with the same responsibility to avoid passing judgment on the author or publisher before a final determination is made.</p><p>Media, especially sensational media, spreads rapidly. This spread is across many platforms, with clips of popular videos being copied from Youtube to Instagram to TikTok to Facebook. Media platforms have an obligation to coordinate mitigation efforts, such that claimants don’t bear the burden of chasing down harmful media across dozens of platforms. There may be regulatory amendments required to simplify such coordination.</p><h3>7. Deterrents</h3><p>The larger harms we are discussing (Foreign interference in elections, Libel, Non-consensual pornography, Fraud) are all serious crimes in their own right; additional deterrents for the use of deepfakes in committing these crimes would likely be irrelevant. What remains, therefore, is the tricky question of “unlabelled political commentary” — or misinformation campaigns masquerading as satire. Making the technicians, consultants and media creatives jointly responsible for the misuse of synthesis technologies is likely the most effective deterrent. Penalties for repeat offenders should be magnified.</p><h3>Conclusions</h3><p>While the threats posed by deepfakes may <em>feel</em> new and emergent, the underlying problem of synthetic media has been accelerating for decades. An attempt to address the harms of such media is long overdue. This framework is a draft, intended to address some gaps and spur further discussion. Feel free to jump in!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=532d2d4b7791" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Case for Going Beyond Chain-of-Custody]]></title>
            <link>https://medium.com/@jmckenty/the-case-for-going-beyond-chain-of-custody-ed5d54b3d241?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/ed5d54b3d241</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[policy]]></category>
            <category><![CDATA[observation]]></category>
            <category><![CDATA[medium]]></category>
            <category><![CDATA[deepfakes]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Fri, 01 Dec 2023 00:44:50 GMT</pubDate>
            <atom:updated>2023-12-01T00:52:20.036Z</atom:updated>
            <content:encoded><![CDATA[<h3>Why Media Verification Requires Federated Observations</h3><p><em>This is part three in a series of posts addressing the requirements of a system for global media verification. </em><a href="https://medium.com/@jmckenty/the-case-for-modular-media-verification-bf93e327bb40"><em>Part Two, The Case for Modular Media Verification</em></a><em>, discussed why our system must be adaptable to the needs of several different user communities.</em></p><p>In the summer of 2019 the President of Gabon, Ali Bongo, had fallen ill. By the Fall, rumors had begun to circulate of his death. He had not been seen in public in months, but the government (a long-running family dynasty) insisted he was alive and well. His New Year’s video address was released, and immediately there was chaos — was it a fake? Or did it simply show the symptoms of a recent stroke? The President’s opponent, Bruno Ben Moubamba, claimed the video was a deepfake, but attempts to prove this were inconclusive. A week later, there was <a href="https://www.motherjones.com/politics/2019/03/deepfake-gabon-ali-bongo/">an attempted military coup</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_65AXI8SiBdG5CuV61GC5g.png" /><figcaption>Justice is Blind</figcaption></figure><p>In order for verified media to useful in diplomacy, we need <strong>symmetry of confidence: </strong>every stakeholder must be able to rely equally on the accuracy and impartiality of the verification system. And when the impacts of such diplomacy cannot be higher, that confidence cannot be based solely on trust in the system operators.</p><h3>Trust, but Verify.</h3><p>Imagine a video of the Canadian Prime Minister, Justin Trudeau, calling for the surrender of Ukraine to the Russians. If such a video was stamped and signed as verified by <strong>Russia Today</strong>, would that provide much assurance to American allies? What if it was a call by President Biden for Taiwan to surrender to China — verified by <strong>Huawei</strong>?</p><p>Equally, there is no way to expect our allies or opponents around the globe to accept <a href="https://c2pa.org/">the verifications of <strong>Adobe</strong> or <strong>Microsoft</strong></a> at face value. In a global context, the host nation of any centralized system has an asymmetric opportunity to tamper with it.</p><p>All parties must have equal access to the mechanics of verification. This rules out systems that are based on provenance assertions provided by standalone entities, no matter how noble or well established. Even if we ignore wiretapping laws, national security imperatives and the potentially-questionable commitment to impartiality by the company’s executive team, <a href="https://www.wired.com/story/the-full-story-of-the-stunning-rsa-hack-can-finally-be-told/">as we learned from RSA</a>, sooner or later all such trust is misplaced.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*8-FRRBds5SDXvc83" /><figcaption>Independent Observations</figcaption></figure><h3>Tamper-Proofing through Observations of Change</h3><p>When kidnappers contact you with a demand for ransom, the first thing any hostage negotiator will do is ask for a “Proof of Life”. Typically this is a photograph of the hostage, holding up today’s newspaper. While there’s no way to know ahead of time what tomorrow’s newspaper will look like, anyone in the world can verify what today’s newspaper looks like — so this ensures that your hostage is still alive; or at least they were at some point today.</p><p>Media verification requires the same thing — except that we need a unique newspaper cover for every frame of video (roughly every 30 milliseconds). These proverbial “newspapers” can be any unpredictable phenomena (weather, celestial events, even stock market prices) that can be independently observed by “enough” of the key stakeholders.</p><p>Much like photographs themselves give us the opportunity to “trust our own senses”, using independent observation as the basis for media verification will allow all parties to trust those mechanics.</p><h3>Put a pin in it</h3><p>Having established a “root of trust” in independent observation, we have avoided the trap of expecting our counterparties to trust our media chain-of-custody. What is left is to commit to immutability — providing a public guarantee that the media in question has not been edited since the verification was established. This is the only part of media verification for which a decentralized ledger is necessary (or even relevant). Any immutable ledger will do for this purpose, but preferably one with a high frequency of settlement. (The keyless blockchain developed by the Estonians and commercialized as GuardTime a year before the release of BitCoin would be adequate, for example).</p><p>What is left, now, is the “watermark” — how can we take critical metadata (the metaphorical newspaper cover, as well as the “address” of the camera) and inject it… not into the digital file (which could easily be done to pre-generated content), but<em> into reality itself</em>, at the moment of capture?</p><p><em>Stay tuned for part four.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ed5d54b3d241" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Case for Modular Media Verification]]></title>
            <link>https://medium.com/@jmckenty/the-case-for-modular-media-verification-bf93e327bb40?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/bf93e327bb40</guid>
            <category><![CDATA[deepfakes]]></category>
            <category><![CDATA[medium]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[policy]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Tue, 28 Nov 2023 06:38:25 GMT</pubDate>
            <atom:updated>2023-12-01T06:49:58.267Z</atom:updated>
            <content:encoded><![CDATA[<p><em>This is the second in a series of posts, articulating the need for a global media verification system, and the features such a system must have. If you missed </em><a href="https://medium.com/@jmckenty/the-case-for-verified-media-3f0f5bb83fd1"><em>post one, </em>The Case for Verified Media</a>,<em> you may want to go read that now. Join the discussion in the comments.</em></p><p>When it comes to matters of fact, there is no utility in having independent sets of them. If Americans and Russians, Democrats and Republicans, or Martians and Earthlings have separate systems for documenting <strong>Reality, </strong>shared trust in those systems is likely to collapse at the moment we need shared reality the most — during crisis. So our goal must be a universally trusted system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ufFg3bgxC8zV2US14_WN_A.jpeg" /><figcaption>Illustration of the Christmas Truce, 1914</figcaption></figure><p>If we are to achieve this goal, the approach we adopt must meet the needs of both our most <em>ubiquitous</em> users, and our most <em>extreme</em> ones.</p><p>By far, the most ubiquitous form of media capture in the world is the smartphone — accounting for 92.5% of all photos on a daily basis. It provides all of the technology needed to verify the location and authorship of the media, and the typical consumer is more concerned with <em>credit</em> (ie establishing copyright and ownership) than <em>privacy</em> (protecting the details of their location and identity).</p><p>However, the requirements of the most <em>extreme</em> media are the opposite — collection of intelligence or documenting the realities of law enforcement, war, or sensitive communications requires the utmost in discretion. The very fact that media was captured must be deniable. Any and all metadata that could indicate the sources or methods of the media, while vital to be able to verify, must be stored independent from the media itself. Publishing of either the media or the metadata cannot be a requirement of verification.</p><p>Thus, our system of media verification must be <strong>adaptable</strong>: allowing us to address the challenges of media copyright, the provenance of related and derived media products, and the tertiary concerns of AI training data and models, <strong>separately</strong>.</p><p>Such a modular system is within our grasp.</p><p><em>Read on to </em><a href="https://medium.com/@jmckenty/the-case-for-going-beyond-chain-of-custody-ed5d54b3d241"><em>Part Three: Beyond Chain-of-Custody</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bf93e327bb40" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Case for Verified Media]]></title>
            <link>https://medium.com/@jmckenty/the-case-for-verified-media-3f0f5bb83fd1?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/3f0f5bb83fd1</guid>
            <category><![CDATA[policy]]></category>
            <category><![CDATA[disinformation]]></category>
            <category><![CDATA[deepfakes]]></category>
            <category><![CDATA[medium]]></category>
            <category><![CDATA[verification]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Wed, 22 Nov 2023 00:05:21 GMT</pubDate>
            <atom:updated>2023-12-01T06:46:43.251Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1ZzlZkQCZtHd9ixwH3xHmg.jpeg" /><figcaption>Real photograph, fake story.</figcaption></figure><p>Since the invention of the camera, <a href="https://dl.acm.org/doi/abs/10.1145/506443.506578">humans have been taking advantage</a> of the fact that we respond viscerally to media. Put simply: Pics, or it didn’t happen. (By which we mean, <a href="https://doi.org/10.1186/s41235-017-0067-2">if pics, then it definitely DID happen</a>). Not only is a picture worth a thousands words, but our brains are hardwired to trust our own senses. (More importantly, they can’t be convinced otherwise, even when we KNOW we’ve been fooled).</p><p>The power of the computer, especially since the introduction of <a href="https://www.frontiersin.org/articles/10.3389/fpsyg.2021.654474/full">ubiquitous digital cameras</a> and the rise of Adobe Photoshop, has allowed us to subvert that implicit trust. First, it was practical effects and then CGI in movies; more recently, has been the rise of the <strong>deepfake</strong>.</p><p>Synthetic media is now a staple of both crime and warfare around the globe: Insurance fraud, identity theft, disinformation campaigns, revenge pornography, defamation, and more: the underlying weapon is the ability for us to convince one another that a fake, isn’t. More insidious still, every new deepfake causes a second harm: it erodes our confidence in the evidence of our senses. This is called the “<a href="https://en.wiktionary.org/wiki/liar%27s_dividend">Liar’s Dividend</a>”, and it arms despots and dictators with a get-out-of-jail-free card. “That was never me in that video,” they cry. “It must be a fake.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QWJO4P-0-_v64e_VLnazkg.jpeg" /></figure><p>Current efforts in the fight against synthetic media focus on the problem of detection — can we analyze a photo or video and tell if it’s a genuine article? This approach has served us well to date, but as deepfakes become the prevailing approach to producing synthetic media, detection is no longer enough.</p><p>Firstly, detection will never been 100% accurate. <a href="https://futurism.com/darpa-68-million-technology-deepfakes">DARPA has funded tens of millions of dollars of work</a> in this area, yet each generation of detection tech becomes less reliable over time. Detection methods can’t keep pace with advancements in generation approaches, leading to two different problems: false negatives, and false positives.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SwBx0gdFI8VrfBRyHibrMQ.jpeg" /><figcaption>Which is Worse: False Positive, or False Negative?</figcaption></figure><p>False positives are an administrative problem for media platforms such as Youtube, Instagram, or Whatsapp, and a resource problem for law enforcement. In the worst cases, media that is wrongly labelled as false can trigger a backlash of censorship claims and “Deep State” conspiracy theories. Even in the best case, a system with a high rate of false positives will quickly be ignored. (Think of an insurance company where 5–10% of car accident photos are wrongly flagged as “synthetic”).</p><p>But false negatives are even worse. Imagine what would have happened if <a href="https://www.npr.org/2022/03/16/1087062648/deepfake-video-zelenskyy-experts-war-manipulation-ukraine-russia">the deepfake video of Zelensky’s surrender to Putin</a> was wrongly flagged as “Not Fake”?</p><p>Given how deepfakes are actually generated, there’s a third problem with detection — every system that is built to detect deepfakes can immediately be weaponized to MAKE deepfakes that avoid this system. (<a href="https://en.wikipedia.org/wiki/Generative_adversarial_network">The underlying technology is called GANs</a>, and its one of the most powerful forms of AI developed in recent years).</p><p>We can avoid this third problem by keeping the underlying mechanisms of detection secret; while DARPA made the mistake of publishing <a href="https://github.com/mediaforensics">their earlier</a> <a href="https://github.com/NVlabs/stylegan3-detector/">detection work</a> as open source, to their credit most commercial detection companies take this approach. But this introduces an additional problem — naysayers won’t trust a detection mechanism that isn’t exposed to public scrutiny, and we’re back to the Liar’s Dividend again.</p><p>The solution to this problem is simple to define, and hard to achieve: We need a global standard for “authentic” media. That authenticity must be established at the moment of capture — the click of the proverbial shutter. And it must be tamperproof — based on mathematics, not on trusted partners.</p><p><em>In the follow-up posts, we’ll address the properties such a system of media verification must have, and the policy implications. Read on with </em><a href="https://medium.com/@jmckenty/the-case-for-modular-media-verification-bf93e327bb40"><em>Part Two: The Case For Modular Media Verification</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3f0f5bb83fd1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Startup Financial Model]]></title>
            <link>https://medium.com/@jmckenty/the-startup-financial-model-7a6b00cbb659?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/7a6b00cbb659</guid>
            <category><![CDATA[startup-finance]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[fundraising]]></category>
            <category><![CDATA[pitch-deck]]></category>
            <category><![CDATA[excel]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Sat, 22 Jul 2023 22:49:40 GMT</pubDate>
            <atom:updated>2023-07-22T23:54:29.207Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*gx8_C7_pkbwaeyKe.png" /><figcaption>Worst financial slide we’ve ever seen.</figcaption></figure><p>Aka the ProForma, the “Numbers Slide”, or the “Plan”</p><p>Along with the ubiquitous <a href="https://en.wikipedia.org/wiki/Product/market_fit">Product/Market Fit</a> conversation, there’s also the less-well-known concept of <a href="https://paulasel.medium.com/founder-market-fit-2167a2536195">Founder/Market Fit</a>. This is a useful way to talk about how folks tend to have the sort of UNIQUE INSIGHTS that translate into long-term success, when they start companies within a market that they already understand.</p><p>The sad side effect of this, however, is that most first-time founders DON’T have a background in finance; for many of them, putting together their first investor pitch deck is the first time they’ve ever had to attack any sort of financial model.</p><p>I’m not an accountant, nor do I have an MBA. But I have been building, scaling, and selling my own startups for over 20 years, and I’ve been helping others on that path for almost as long. So at least once a month, I end up needing to help a first-time CEO build out, tweak, and tune their “Model”. <a href="https://delvinghq-my.sharepoint.com/:x:/g/personal/joshua_mckenty_delvinghq_onmicrosoft_com/EZbXA2LQ8DpGpeCR7ELrz2UBJHUegrAdPi576stUVMKOSQ?e=puOzgh">Here is my model</a>, and here are my tricks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*XVPMMJ3TrkSOpHEaUdr6MQ.png" /><figcaption>These days, every model needs to end in a profitable state.</figcaption></figure><h3>Variable start date</h3><p>The rare, lucky and privileged entrepreneur has enough personal savings to start their company <em>today</em>, and grow at an ideal speed regardless of outside investment. But for the rest of us, the “start date” of our business is heavily influenced by when we successfully complete fundraising. Making this a single variable takes some Excel magic, but it’s worth it.</p><h3>Staffing costs using FCA</h3><p>While it’s not perfectly accurate in the long term, for startup models I like to use “Full Cost Accounting” attached to headcount. This means I factor ALL of the overhead costs of the business (office space, equipment, internet, payroll taxes and benefits) into the salary estimates, and then I carefully project when we’ll hire, and how many of each employee type. For a remote team of less than 10 people, I find 35% overhead to be pretty accurate.</p><p>Staffing levels are the only thing I model manually on a month-by-month basis (without some sort of automated “ramp”), as I think it’s important to be able to tell the story of each hire in an early team.</p><h3>Ramped values</h3><p>First-time spreadsheet jockeys will often hardcode a given amount of revenue for each month. More experienced executives will use a monthly “growth rate” on one or more of their key business drivers (number of leads, conversion rate, average deal size, etc). But neither of these really match how we think and talk about our business.</p><p>Instead, I use three fields for each key variable: a “start” and “end” field, and a “ramp” field. This allows me to model change of these key parameters over time, in the way I usually end of talking about it with my board members and professional advisors. E.g., “We’re going to start the period with 30% of our revenue coming from consumer accounts, but we expect to drive that down to 15% at a linear rate over three years.” Or, “Our Enterprise accounts have an average contract value of $50,000 per year; we expect to grow that to $150,000 per year, but most of that improvement will happen in the end of the period.”</p><p>Being able to “tune” these sort of ramped business drivers really makes it simpler to dive into the “HOW and WHY” of the business — and seeing all these assumptions on the front page of the model fosters great conversations with investors.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IaZ_sp5CNtR-1wGbV31YBg.png" /><figcaption>Variables with start, end and ramp. (Linear ramp is 1; Cubic ramp is 3).</figcaption></figure><h3>Conversion funnels and Churn</h3><p>Every SaaS founder knows that they’re going to live or die based on customer churn; getting that churn reflected in their Income statement and active account count is harder than it looks, especially when you want to use a ramped and variable churn rate. These days I always model a Lead-&gt;Pilot-&gt;Paid-Account-&gt;Retained-Account flow, even if I’m assuming those deals will be sales-led and not PLG or marketing-driven. Spending more money to get more leads (while acknowledging that the cost-per-lead is probably going UP at scale) is the simplest way to get a reasonable view of your business’s potential velocity.</p><h3>Date-based spend</h3><p>We’ve always got some “big ticket items” in mind for the budget — often that’s pulling the trigger on PR spending, or adding a key outside contractor for compliance work. I like to set up those expenses with an adjustable “start date”. When that one investor says “How can you go faster?”, you’ve got an answer waiting.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NR9ldLVUDBqcxy3C0tod2g.png" /><figcaption>Monthly expenses with variable start-dates</figcaption></figure><h3>Rolled-up Quarters and Years</h3><p>There’s no way to accurately model a financial forecast without doing it month-by-month, but no one actually wants to have to read it that way. I’ve got some cute-but-trivial date math that makes it easier to see rolled-up totals at the Quarter and Year level. Make sure you notice when you’re looking at partial years — unless you start on January 1st, both the first and last year of your model will be incomplete.</p><h3>Copy-Pasta-Ready Table and Graph</h3><p>To make sure that we avoid lumpiness from the partial years, the graph is driven by the raw monthly data. (This also makes it easier to make sure your bank balance never drops below zero!)</p><p>The final output of this model is almost always an Excel slide, so I’ve got a well-formatted (rounded-down) table on the last page for copying and pasteing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1006/1*6MgZh1ieR5L8O0J5Jf65sA.png" /><figcaption>The final table is “ready to copy” into Powerpoint.</figcaption></figure><h3>Disclaimers and further work</h3><p>No model is perfect, nor should it be. This model doesn’t include any non-product revenue (attached services for Enterprise deals, training and support, etc), and it doesn’t separate marketing and sales costs very well. There are MANY other expenses that become significant over time, which don’t fit into the FCA-based approximation. For simplicity, we’re applying a single conversion rate and a single churn rate to three different product tiers. We don’t model account expansion over time, and the connection between ISR headcount and deal close rate is completely manual (as is the ramp-up time for new reps). Finally, the dollars in this example are completely fictional — they’re an amalgam of US and Canadian labour rates, and example SaaS product price points.</p><p><strong><em>Have you got a different or better approach to Startup financial modelling? Share your tips and tricks in the comments!</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7a6b00cbb659" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to code properly — Part 1: Run]]></title>
            <link>https://medium.com/@jmckenty/how-to-code-properly-part-1-run-17f2eca502eb?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/17f2eca502eb</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[ssl-certificate]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[learning-to-code]]></category>
            <category><![CDATA[web-app-development]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Wed, 24 May 2023 22:56:11 GMT</pubDate>
            <atom:updated>2023-05-24T22:56:11.293Z</atom:updated>
            <content:encoded><![CDATA[<h3>How to code properly — Part 1: Run</h3><figure><img alt="Green text of the word “HELP” on a black background, visible on the screen of an antique Apple II+ computer that’s rusty and surrounded by leaves." src="https://cdn-images-1.medium.com/max/640/1*c5LEO14mmf_RJ_enCeGrVQ.jpeg" /></figure><p>My first programming language was Apple Basic. It was 1984, and the computer was an APCO — the very first Apple II+ <em>clone</em>. Like the real Apple II+, if you turned it on without anything in the floppy drive, you would arrive at a command prompt that was also the BASIC interpreter.</p><p>Any short basic program could be entered simply by typing lines of code directly into that command prompt — no editor required. And then… you simply typed run.</p><p>These days, it’s common for developers to work for weeks, months, or sometimes even YEARS before their code makes its way to “Production”. And production might look so different from their local development environment, it’s hard to imagine that code could work at all. This has led to one of the canonical axioms of modern development: <a href="https://12factor.net/dev-prod-parity">Dev/Prod Parity</a>.</p><p>The principle of Dev/Prod Parity is simple: make your development environment as similar to Prod as possible, so that you won’t end up with a bunch of bugs that are directly related to the differences. But in a world where “<a href="https://www.youtube.com/watch?v=cBhZ6S0PFCY">HTTPS Everywhere</a>” has become table stakes, most new developers are confronted with a terrible obstacle: before they can have that magical run experience, they have to turn their local machine into a production server. This means…</p><ol><li>Buy a domain name</li><li>Configure DNS records for that domain name to point to their local machine</li><li>Install <a href="https://certbot.eff.org/">certbot</a>, and use it to set up certificates for those names</li><li>Install NGINX or Apache, and configure it to use those certificates</li><li>Then, configure NGINX to “reverse proxy” to the port where their local application is <em>actually</em> running</li></ol><p>Every time you add a new application or service to your project, you’ll need to do steps 2–5 over again. And each time, you risk breaking the configuration that worked for the LAST service.</p><p>Back in 1984, there was no internet. So when I felt proud of my programming, I would ask my mom (or my sisters, or the neighbor) to come and look over my shoulder at the green and black screen. But today, when I want to share what I’m proud of, I have a few different options:</p><ol><li>I can put it into “staging” — yet another copy of “production”, with the same maintenance burden and negligible costs</li><li>I can host a Zoom call (or some other screen-sharing video conference) and walk folks through it (but they won’t be able to try it for themselves)</li><li>I can set up a “Reverse SSH Tunnel” using a service such as NGROK — as long as I can get it configured and running</li></ol><p>These two activities, run and share, are the first two steps of becoming a programmer. They set up a feedback loop that accelerates learning. Yet, in the nearly 40 years since <strong><em>I</em></strong> started, they’ve gotten <strong><em>harder</em></strong> rather than easier.</p><p>I’m working on changing this. (And yes, this is <a href="https://tanzu.vmware.com/content/blog/co-founder-of-piston-openstack-board-member-joshua-mckenty-joins-pivotal">the same problem I joined Pivotal to work on</a> back in 2014). Today’s effort is called <a href="https://loophost.dev/">LoopHost</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MjmfYgoYJWvJZSSxIqkNOg.png" /></figure><p>LoopHost is an <a href="https://github.com/delving-co/loophost">open source library</a>, written in Python and Go, that wraps up and simplifies all of the steps described above. It’s also the web service that makes the DNS and tunneling parts seamless.</p><p>Installation is as close to the zero-click experience from the 80s as we could get — it’s a single command, typed into the terminal. After that, everything in loophost happens in the browser.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XBLnDlfJ5zAujoSrZ2ASoQ.png" /></figure><p>We need to reclaim the “build, run, share” loop of joy. Please join me.</p><p>]run</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=17f2eca502eb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Staying in Flow]]></title>
            <link>https://medium.com/@jmckenty/staying-in-flow-e4347186be7f?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/e4347186be7f</guid>
            <category><![CDATA[flow]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[productivity]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Mon, 08 May 2023 23:19:23 GMT</pubDate>
            <atom:updated>2023-05-08T23:19:23.352Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="abstract white and blue lines expressing flow state" src="https://cdn-images-1.medium.com/max/768/1*kJMbeklIFSijF2MrIm2Bgw.png" /></figure><h3>(…<em>or</em>, Why I’ve Become Obsessed with Single-File Programming and You Should, Too)</h3><p>The big lie we’re told, in school and again in early-career settings, is that being “good” at programming is about solving harder and harder problems in a more elegant way. The truth is much more mundane — being good at programming is about efficiently solving a large set of mostly trivial problems, quickly and cleanly. What IS true, however, is how valuable “flow state” is to doing this. Being able to remain immersed in the work allows you to quickly shift between domains, tracking the flow of state from user to business logic through to data store, and back again.</p><p>Many “Good” programming practices add complexity:</p><ol><li>Separate models and views and controllers.</li><li>Keep all your tests in parallel files and folders.</li><li>Don’t inline CSS — keep it in a hierarchy of separate CSS files, referenced by class and ID.</li><li>Don’t even WRITE CSS — use a preprocessor and generate it from some intermediate language.</li><li>Use nested namespaces of packages and modules, each with their own Standard conVentions for nam-ing thin_gs.</li></ol><p>Demanding this sort of early rigor is not only pointless gatekeeping; it’s probably making all of us less efficient.</p><p>In much the same way that survey scales are plotted against the average human’s ability to discriminate (usually 5-7 choices, never more than 9), frameworks that rely on fewer files, of fewer types, and organized in as flat a hierarchy as possible are simpler for new developers to come up to speed with, and easier for EVERYONE to remain in flow state when working on.</p><p>“Good” programmers love to denigrate PHP and javascript for being “bad” languages; yet both of them are notable for supporting simple “Single-file” patterns. Javascript does this through JSX; PHP simply allows every php script to contain arbitrary HTML as well. I’ve decided to recreate the joy of Single-File development, but for Python. This will hopefully bridge the obvious gaps between the “Notebook” python data science communities, the “Boot Camp” software developer communities, and the front lines of web application development.</p><p>Over the next few blog posts, I’ll talk through many of the “bad” practices I’ve embraced to simplify and streamline my development experience. Even better, I’ll announce the various open source libraries that I’ve discovered, extended or developed along the way.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e4347186be7f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When Clothes Overwhelm: Understanding Sensory Overload in Children and How to Help]]></title>
            <link>https://medium.com/s-m-ooth/when-clothes-overwhelm-understanding-sensory-overload-in-children-and-how-to-help-b30f42e94c6?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/b30f42e94c6</guid>
            <category><![CDATA[clothing]]></category>
            <category><![CDATA[neurodiversity]]></category>
            <category><![CDATA[highly-sensitive-people]]></category>
            <category><![CDATA[parenting]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Mon, 03 Apr 2023 19:13:11 GMT</pubDate>
            <atom:updated>2023-04-03T19:13:11.227Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Very upset young child" src="https://cdn-images-1.medium.com/max/1024/1*19GYtcmHQcpFLLpLPgIPyQ.png" /></figure><p>Clothes can be overwhelming to children (and adults!) with sensory processing issues. As a father of neuro-diverse children (and someone who struggles with sensory processing myself), I know this all too well. When a child becomes overwhelmed by their clothes, it can trigger a sensory overload response in their brain. Seemingly out of nowhere, your child may start to scream, cry or get angry.</p><p>For some children, certain fabrics or textures can be especially triggering. The sensation of scratchy clothing, for example, may cause them to feel uncomfortable or even in pain. Similarly, tags, seams, or other features of clothing can create pressure or friction against their skin. Sensitivity can be worse in hot weather, or when it’s unusually dry or humid. And children can be more sensitive to their clothing when they’re overwhelmed by seemingly unrelated stimuli, such as background noise, allergies, indigestion, or even flickering lights.</p><p>When a child experiences sensory overload, their brain may respond with a fight-or-flight response. This can cause them to feel anxious, irritable, and physically uncomfortable. You may notice your child becoming agitated, fidgety, or experiencing a meltdown.</p><p>It’s important for parents to understand that this response is not the child’s fault, and it’s not a behavior that they can simply “snap out of”. Rather, it’s a real physiological response that is beyond their control.</p><p>Fortunately, there are steps that parents can take to help their children manage sensory overload. This includes finding clothing that is made from soft, comfortable materials, or choosing clothes that don’t have tags, exposed zippers, or other features that could cause discomfort. It may also involve working with a therapist or occupational therapist who can help the child develop strategies for managing sensory overload in a variety of settings.</p><p>Ultimately, the key to helping a child manage sensory overload is to be patient, supportive, and understanding. By working together to find solutions and provide a safe, supportive environment, parents can help their children thrive and reach their full potential.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b30f42e94c6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/s-m-ooth/when-clothes-overwhelm-understanding-sensory-overload-in-children-and-how-to-help-b30f42e94c6">When Clothes Overwhelm: Understanding Sensory Overload in Children and How to Help</a> was originally published in <a href="https://medium.com/s-m-ooth">s(m)ooth</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hijacking the Moment of Intent]]></title>
            <link>https://medium.com/@jmckenty/hijacking-the-moment-of-intent-863b3e180aaf?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/863b3e180aaf</guid>
            <category><![CDATA[buy-black]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[shopping]]></category>
            <category><![CDATA[ecommerce]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Mon, 21 Sep 2020 22:32:20 GMT</pubDate>
            <atom:updated>2020-09-21T22:32:20.738Z</atom:updated>
            <content:encoded><![CDATA[<p>For decades, investors and entrepreneurs have been obsessed with capturing the “Moment of Intent”. This drove the early days of the internet with increasing refinement of search engines, the development of cross-sell and up-sell recommendation engines, and an explosion of retargeting and other data-driven approaches to advertising.</p><p>But what about values?</p><p>In the heat of a dopamine-fueled impulse buy, do our values get a chance to speak? Or is the craving for instant-gratification and overnight shipping too strong a siren call?</p><p>I’ve decided to find out (and hopefully, do some good in the process). I’ve built a new free tool, called the <a href="https://buyblackbutton.com/">“Buy Black Button”</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/534/1*1SpH-1GM378UKTzWAkH1Kw.png" /></figure><p>BuyBlack is very similar to a tool I built 13 years ago, called “BuyLatr”. It, too, added a button to each Amazon product page. In the old world, we detected price changes and let you know when the product went in or out of stock. (Perfect for picking up a Wii as a holiday gift.) In this world, we give you a chance to look at a similar product, from a Black-Owned Business.</p><p>I’m a huge fan of ubiquitous computing. I like to be reminded to do what I said I was going to do, and there’s no better time than when I’m about to click “Buy Now”.</p><p>What do you think?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=863b3e180aaf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Dear Graduates, You’re Not “Junior Developers”]]></title>
            <link>https://medium.com/built-to-adapt/dear-graduates-youre-not-junior-developers-53b80e020713?source=rss-aa6edb551662------2</link>
            <guid isPermaLink="false">https://medium.com/p/53b80e020713</guid>
            <category><![CDATA[developer]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[graduation]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Joshua McKenty]]></dc:creator>
            <pubDate>Wed, 20 Jun 2018 18:06:00 GMT</pubDate>
            <atom:updated>2018-06-26T20:15:45.437Z</atom:updated>
            <content:encoded><![CDATA[<h4>Why you shouldn’t think of yourself as a junior developer</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9eYMpBb7ZonEudY1KjJhbA.png" /></figure><p>A lot of young people will graduate this month and start looking for a job. They’re polishing their LinkedIn profiles, deleting those beer-bash photos from their social media accounts, and deciding on a career path they think will bring success. In a way, entering the workforce is a bit like attending a Halloween party: “What should I go as?”</p><p>If you’re entering the tech workforce, you might be wondering if you should go as a “junior developer.” Well, there is a lively debate online right now about this very topic.</p><p>One side says emphatically no, you should not. <a href="https://hackernoon.com/dont-be-a-junior-developer-608c255b3056">“Seriously, don’t be a junior developer.”</a> This side says, you should learn to code on your own. Do some side projects and don’t enter the workforce until you’re at least an intermediate coder or a solid plain coder with nothing junior about you.</p><p>The other side says yes, go ahead, be a junior developer. Although they admit it won’t be easy.<a href="https://medium.com/@melissamcewen/who-killed-the-junior-developer-33e9da2dc58c"> “For the junior devs out there, it’s going to be hard to land your first job,”</a> they concede, but it’s not your fault. The industry is to blame for the shortage of junior developer jobs. Companies think it takes so much time and money to train junior developers that they only list positions for senior devs these days.</p><p>Before we go any further on this topic, let’s take a step back — way back. Let’s first talk about the society we live in and let’s start with the role that privilege plays. Some people are bestowed with the privilege of being hired as coders — in fact, I’ve argued publicly with Google about its hiring process. Some people also have the privilege of being well off. They can afford to be unemployed long enough to interview wherever they want, rather than take the first job that’s offered. This privilege of being able to coast doesn’t get talked about enough.</p><p>A corollary of coasting is the side project. “Do a side project and improve your skills,” says the no-on-junior-developers faction. But side projects take time (often free time), and time is not a privilege many people have. Not everyone can live in their parents’ basement while they work on a side project. Like me, for example. My family didn’t even have a basement. Actually, I take that back. We did. We lived in a church basement when I was a kid.</p><p>What I’m saying here is that economic privilege is a real issue when it comes to the whole concept of junior developer. Although, if you are bestowed with privilege, there’s nothing wrong with that. In fact, maybe you owe it to your parents — or your grandparents or whoever — to put your privilege to good use.</p><p>I don’t like the term “junior developer.” I think we should title such positions “apprentice developer,” because it’s not about the quality of the developer’s output — it’s about the amount of supervision they require.</p><p>You can have a senior developer who’s lazy or not detailed-oriented or undisciplined. One of the smartest coders I ever worked with, if you didn’t supervise him he’d go off and write a compiler.</p><p>He actually did on one project. We were trying to get him to put together a scripting interface, but he wrote an entire language and a compiler for it instead of the browser plug-in we needed him to build. So in my book, he was a very senior coder but still an apprentice because he needed so much supervision.</p><p>But there are other reasons to stop using the words “junior” and “senior.” For one, they imply old and young. For another, they imply male. In a family, women are never junior or senior. The titles “junior developer” and “senior developer” indicate on their face that we’re talking about men, and there is undeniable gender bias attached to them.</p><p>By contrast, “apprentice” is a meaningful title — for the developer and the company. And if the company has carrying capacity, it’s a great way to grow talent. Although, proviso: the apprentice has to be apprenticing in skills they can learn, not skills they’re never going to master. Some minimum level of aptitude is required.</p><p>And now that I’ve provided my context for the “junior developer” discussion, let me start by saying to those of you graduating now: my advice to you is to ignore the interview process and go after the job you want. When I came out of school (Maxwell International High School in Shawnigan Lake, British Columbia) I decided where I wanted to work, went there, and told them I was going to work there.</p><p>Remember that the skillsets you need will change. It’s actually easier to follow smart people into new fields than to chase fields. If you have privilege and you can go chase what you want, decide literally where you want to work (technology, geography, and the sign on the door) and then figure out what you could do for them that would make a difference.</p><p>I have done this. I took a one-way bus trip to rural Iowa, showed up at this gaming startup, and just started fixing stuff for them. At the end of my first day, the CEO offered me a job. I showed them not only could I do the work, I could figure out what work needs to be done, which is always the higher-value item.</p><p>I learned a lot there. But most young people don’t choose the job where they’re going to learn the most. They choose their job according to what it pays, or maybe where it’s located (i.e. not Iowa). That’s totally fine — but realize that sometimes the job that pays more won’t be where you learn more — and if you’re okay with that decision that’s great. But every time I had a choice between making more money or learning more, I opted for learning more. I was always willing to leave quickly and do things I was terrified of — and I’d encourage you to do the same if your goal is to learn. Most folks don’t, because this is irrational — but in the context of coding, it isn’t.</p><p>Again, don’t think of yourself as a junior developer; think of yourself as an apprentice. And you’re apprenticing for more optionality in the future so you can make better money and work wherever you want.</p><p>To be a successful apprentice, you should learn how to self-educate and how to ask for help. I have a friend who was an acupuncturist, and he decided he wanted to be a coder. At the time, he had four friends who were coders and, when working through a tutorial, he’d need to pose some basic questions. But he made sure to spread his questions among the four of us, so as not to overly pester any one person. He was willing to self-educate and work through the tutorial, but he was also strategic about asking for help when he got stuck.</p><p>That’s the same approach you should take when you’re apprenticing in a workplace. Try to do it on your own, but ask questions if you’re stuck for more than five minutes. Don’t spend a day being stuck, because that’s a waste of time for you and your employer. Don’t pester people, but find that middle ground. Make sure your questions get better over time, and that you’re not asking the same questions over and over again. That’s annoying.</p><p>If you take away only one thing from this piece, I hope it’s this: there’s no such thing as a “junior developer,” there’s apprenticeships — and either you’re either apprenticing towards a lifestyle you want or you’re apprenticing to position yourself for a future opportunity. Neither choice is wrong or right, just decide for yourself which is yours and don’t let anyone call you “junior” along the way.</p><figure><a href="http://pivotal.io/"><img alt="" src="https://cdn-images-1.medium.com/max/800/1*tnuuj6HXuZIGsVke00EfbA.png" /></a></figure><p><em>Change is the only constant, so individuals, institutions, and businesses must be </em><a href="http://builttoadapt.io/"><em>Built to Adapt</em></a><em>. At</em> <a href="http://pivotal.io/"><em>Pivotal</em></a><em>, we believe change should be expected, embraced, and incorporated continuously through development and innovation, because good software is never finished.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=53b80e020713" width="1" height="1" alt=""><hr><p><a href="https://medium.com/built-to-adapt/dear-graduates-youre-not-junior-developers-53b80e020713">Dear Graduates, You’re Not “Junior Developers”</a> was originally published in <a href="https://medium.com/built-to-adapt">Built to Adapt</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>