<?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 Rachel Berryman on Medium]]></title>
        <description><![CDATA[Stories by Rachel Berryman on Medium]]></description>
        <link>https://medium.com/@rachelkberryman?source=rss-3f9e7bda6adb------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*gFjNNYyNMVffwhb23Czujg.jpeg</url>
            <title>Stories by Rachel Berryman on Medium</title>
            <link>https://medium.com/@rachelkberryman?source=rss-3f9e7bda6adb------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 17:39:38 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rachelkberryman/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[Why we need human intelligence to unlock artificial intelligence]]></title>
            <link>https://rachelkberryman.medium.com/why-we-need-human-intelligence-to-unlock-artificial-intelligence-21ce0d50cce2?source=rss-3f9e7bda6adb------2</link>
            <guid isPermaLink="false">https://medium.com/p/21ce0d50cce2</guid>
            <category><![CDATA[digitalization]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[management]]></category>
            <category><![CDATA[upskilling]]></category>
            <category><![CDATA[future-of-work]]></category>
            <dc:creator><![CDATA[Rachel Berryman]]></dc:creator>
            <pubDate>Tue, 31 Jan 2023 11:29:49 GMT</pubDate>
            <atom:updated>2023-01-31T11:29:49.495Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZL0nt3fOlWwOoY-j43GeRQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@headwayio?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Headway</a> on <a href="https://unsplash.com/photos/5QgIuuBxKwM?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>Successful AI adoption depends on training people, not on training models.</p><p>Technology is never a complete solution on its own. A new, high-tech advance in orthopedic surgery will not improve a patient’s life quality if the patient skips out on the requisite follow-up physical therapy. Humans are the deciding factor in the success of many technologies, and AI is no different.</p><p>What human factors are currently blocking AI adoption, and how can companies (and the data scientists they employ) keep these humans from dampening the progress AI is capable of bringing?</p><h4>For company executives</h4><p>Managers must first understand what AI really is. Leaders who do not know the difference between machine learning and traditional rules-based models, what a trained model is, or the difference between supervised and unsupervised learning, need to be upskilled before trying to lead a company on its AI journey. These concepts have become as important to understand as cloud vs. on-prem and the basics of cybersecurity. Without a basic understanding of how machine learning really works, companies remain at the mercy of those claiming to know more. Often, this means upper-level management hiring “AI geniuses”, people who often have a PhD in statistics or a relevant STEM field, who are coming out of academia or research. Although such people are undoubtedly quite smart and skilled in their area, they are fundamentally the wrong hire at this initial stage of AI adoption. Hiring them will expose an <a href="https://medium.com/@rachelkberryman/the-data-science-expectations-gap-227610adbc3f">expectations gap</a>: they expect clean data, and freedom to focus primarily on algorithm development, like they had in research. Managers, however, expect revenue-driving machine learning models that quickly bring tangible business value. These expectations are not aligned. At this formative stage, companies must make sure to hire people who are in line with the expectation to bring value quickly using AI, who deeply understand business challenges and are laser-focused on outcomes, not on the process of model training. This profile is called a <a href="https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2022-and-a-half-decade-in-review">translator</a>.</p><p><strong>An AI translator is someone who translates business problems into data science questions</strong>. They have hands-on experience training machine learning models and coding, but are often not the most cutting-edge researchers. At the start of your company’s journey, you do not need cutting-edge researchers. You need to build experience with relevant AI use cases in your company, and skeptical employees and managers will get on board the AI train much more quickly if they are shown quick-win-solutions to their most pressing problems.</p><p>TL:DR for Executives: Educate yourself, then your company. Start with hiring AI translators, not technical experts.</p><h4>For data scientists</h4><p>Data scientists have a big role to play in bringing people along on the AI journey. They need to evangelize AI as a tool that will really help solve peoples’ problems. For too long, data scientists have focused on the technical side of the solution, and on a very narrow part of that technical solution: usually just algorithm training. Data scientists should focus on helping business leaders find relevant and valuable use cases, and on getting models into production as quickly as possible.</p><p>Data scientists also need to be upskilled, for example in the areas of design thinking and user-centric design, and on the software skills designed to support them, like DevOps. Data scientists should learn to apply the MLOps framework to their work. MLOps is a modified form of DevOps that applies its core principles of continuous delivery, quickly incorporating user feedback, and delivering small improvements frequently to ML applications. Using the MLOps framework means deploying models much earlier in the development process than data scientists are typically used to. It means deploying a model that is not perfect, but that allows end users to already test it out and give feedback.</p><p>TL:DR for Data Scientists: Focus on getting models into production. Improve your communication skills. See yourself as a service provider and problem solver (aka AI Translator), not an experimental researcher.</p><h4>Summary</h4><p>Education will be key for scaling AI in the companies and industries where it’s currently lagging. Both executives and data scientists need to speak the same language, that of end-user value. If models are not deployed, they don’t bring value to the end user. If the models don’t answer the right business challenges, they won’t bring value.</p><p>Companies have to put their best human intelligence to the task of AI: not to the task of researching, of building and tuning models, but to tackling the low-hanging-fruit problems and building a solid data and ML infrastructure for deploying trustworthy and scalable models as quickly as possible. For this, <a href="https://engineering.zalando.com/posts/2022/04/zalando-machine-learning-platform.html">Zalando’s ML Platform</a> is a great case study. They invested in a team whose sole purpose is making it easy for data scientists to move models into production.</p><p>Investing in solving tangible business problems with models in production will drive organic growth in AI adoption by removing the human blockers to successful AI implementation.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=21ce0d50cce2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Sapped of Resources — the real source of declining vitality in rich countries]]></title>
            <link>https://rachelkberryman.medium.com/sapped-of-resources-the-real-source-of-declining-vitality-in-rich-countries-e91ef10af8ea?source=rss-3f9e7bda6adb------2</link>
            <guid isPermaLink="false">https://medium.com/p/e91ef10af8ea</guid>
            <category><![CDATA[economics]]></category>
            <category><![CDATA[politics]]></category>
            <category><![CDATA[economy]]></category>
            <category><![CDATA[income-inequality]]></category>
            <category><![CDATA[economist]]></category>
            <dc:creator><![CDATA[Rachel Berryman]]></dc:creator>
            <pubDate>Thu, 29 Dec 2022 12:59:20 GMT</pubDate>
            <atom:updated>2022-12-29T12:59:20.542Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/0*6oJhZGHr7JP7FDVk" /><figcaption>Photo from Morgan Hausel on <a href="https://unsplash.com/photos/PcDGGex9-jA">unsplash</a></figcaption></figure><h3>Sapped of Resources — the real source of declining vitality in rich countries</h3><p>Response to <em>Sapped of Vitality: Why are the rich world’s politicians giving up on economic growth? </em>in The Economist.</p><p>The authors of a recent <a href="https://www.economist.com/leaders/2022/12/14/why-are-the-rich-worlds-politicians-giving-up-on-economic-growth">article</a> in The Economist asked: <em>Why are the rich world’s politicians giving up on economic growth? </em>I ask in response,<em> Why are The Economist’s authors advocating for a focus on outdated measures of wellbeing?</em></p><p>The article calls for governments in rich countries to re-focus their attention on GDP-measured growth, lamenting that the average growth rate of 2.25% seen between 1980 and 2000 has fallen to 1.1% in the years since.</p><p>Having studied economics as an undergraduate (and environmental economics as a graduate) a full 10 years ago, I was shocked to read this recommendation to increase focus on GDP growth in 2022, in The Economist of all places. Already in the early 2010s, there existed a deep well of <a href="https://wcfia.harvard.edu/publications/mismeasuring-our-lives-why-gdp-doesnt-add#:~:text=In%20place%20of%20GDP%2C%20Mismeasuring,issues%2C%20here%20is%20an%20essential">literature</a> about why GDP was a horribly problematic yardstick by which to measure our collective prosperity. Its equation does not factor in environmental damage, taking resources as a limitless input and leaving many externalities out of the equation, to be borne by society. I was surprised and disappointed to see the Economist arguing for a return to this arcane measure of material growth, especially now, when the effects of climate change are showing themselves ever more forcefully. The Economist itself has even published <a href="https://www.economist.com/climate-change">many articles</a> imploring governments to act to stop climate catastrophe.</p><p>Despite this, the article advocates exactly that: a return to focus on the unsustainable, <a href="https://www.overshootday.org/">multiple-planet-consuming</a> GDP growth. The author says that governments are now focusing “too much” attention on aging people, as if the geriatric populations of the US and UK were thriving and flush with cash from over-generous pension payments (in reality, elderly people in both countries <a href="https://www.oecd-ilibrary.org/docserver/factbook-2013-110-en.pdf?expires=1672236078&amp;id=id&amp;accname=guest&amp;checksum=F66B478F6DFFD47038D2D50A043E1019">remain poor</a> compared to other age groups, and the ability to live on pensions and social security is <a href="https://www.ageuk.org.uk/latest-press/articles/2022/life-is-becoming-unaffordable-for-pensioners-on-low-and-modest-incomes-warns-age-uk/">declining</a>).</p><p>In advocating a return to focusing on economic growth, the author seemed dumbfounded that the anti-growth sentiment has “surged since the 80s”, without recognizing the social ills that have accompanied the growth achieved since then. The graph below, from Scientific American, shows how income inequality rose in line with GDP growth in the US.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*kNpcHXbwHeLkGMsB.png" /><figcaption>Source: <a href="https://www.scientificamerican.com/article/gdp-is-the-wrong-tool-for-measuring-what-matters/">https://www.scientificamerican.com/article/gdp-is-the-wrong-tool-for-measuring-what-matters/</a></figcaption></figure><p>The author, presumably an economist, seems also to be unfamiliar with the concept of diminishing marginal returns.</p><p>Indeed, high-growth years like the 50s and 60s saw massive improvements in life quality (although, one could also argue that many of these were not solely attributable to growth, and rather the result of the expanded social safety nets <a href="https://academic.oup.com/book/41850">ushered in</a> by the New Deal order. Correlation with high growth during this time period is not complete causation: there were other factors at play, such as increasing union membership, which <a href="https://scholar.princeton.edu/sites/default/files/kuziemko/files/union_submitted.pdf">drove up wages</a> and decreased income inequality during that time.) Either way, there were massive gains to be had during these times, for example in life expectancy and health outcomes. Since the 80s, we have clearly been in an era of decreasing marginal returns on many of these outcomes, such as life expectancy, where western countries are likely <a href="https://www.weforum.org/agenda/2021/07/how-long-will-people-live-in-the-future/">nearing the top</a> of what is feasible for the human body. Furthermore, there is evidence that growth in rich countries, which up to now has been driven by consumer spending, has <a href="https://www.researchgate.net/publication/223066356_Demand_saturation-creation_and_economic_growth">reached its limits</a> due to saturated consumer markets.</p><p>The author also attributes declines in what they call “growth-drivers” to red tape and rent-seeking behavior.</p><p>Perhaps it’s not “red tape” which mars growth-driving sectors like education and healthcare, but rather privatization, which put profit motive and shareholder value above the purported missions of these institutions and turned poor sick people and first-generation college students into cash cows for the privileged class. These impacts of privitization, as well as the rent-seeking behavior that the author correctly identifies, are also both symptoms of the privatization waves of the late 80s and early 90s.</p><p>“Welfare states have become focused on providing the elderly with pensions and healthcare, rather than investing in growth-boosting infrastructure or the development of young children”, the author says. Here, the author conveniently fails to mention that the welfare state has been <a href="https://en.wikipedia.org/wiki/Welfare_reform">gutted</a> and, since the 80s, has been trying to do more or even the same with much, much less, as wealth was syphoned off to shareholders in privatized industries. To look at where public investment money has gone, one need only look at the precipitous <a href="https://tradingeconomics.com/united-states/stock-market">rise in stock earnings</a> and the gross <a href="https://www.pewresearch.org/social-trends/2020/01/09/trends-in-income-and-wealth-inequality/">increase in wealth inequality</a> between the bottom and top earners in rich nations during this time period.</p><p>To claim that “welfare states”, (as if they even still exist in the US and UK), are failing as a result of catering too much to old people is like claiming that a leaky garden hose with the tap steadily being turned off is failing to keep the lawn green because it’s not being attached to the proper sprinkler.</p><p>The author also fails to address how recent politicians, many of whom were openly committed to GDP growth, actually furthered the declining prosperity problem. For example, the author says Donald Trump “hindered long-term prosperity by undermining the global trading system”. This ignores the fact that he <a href="https://www.newsweek.com/under-donald-trumps-watch-national-debt-increased-78-trillion-1561627">increased national debt</a> by 40% and <a href="https://www.nytimes.com/2020/02/10/business/president-trump-budget-cuts.html">cut support</a> for many “growth-drivers” like education and healthcare, while simultaneously giving even more massive tax breaks to the ultra wealthy. Decrying Trump’s protectionist trade policies as the main reason for declining wellbeing in the US is like blaming a single car driving on a bridge as the cause for its collapse.</p><p>The rebuttals made so far to the author’s arguments have not even addressed the fundamental problem in their advocacy of a focus on GDP growth: that “growth” measured by GDP is completely unsustainable. It is premised the earth’s natural resources being inexhaustible.</p><p>Perhaps politicians are no longer focused purely on economic growth because they are now more focused on preserving the planet for the next generation. The Economist author seems to forget that without a planet to live on, the argument of how to increase growth and prosperity is moot. Rather than preaching continued “economic growth” and a return to that 2.25% average GDP growth, driven by ever-increasing consumerism which pushes up <a href="https://www.overshootday.org/">world overshoot day</a> each year, such an esteemed magazine should focus on new ideas.</p><p>The article ends in the typical way in which many try to encourage policy action at present: by stoking fears about Russia and China (although the inclusion of Russia is confusing, given that they are <a href="https://www.forbes.com/sites/stuartanderson/2022/12/05/russias-economic-prospects-have-gone-from-bad-to-terrible/">not exactly beacons</a> of economic prosperity, and even before the war, had <a href="https://tradingeconomics.com/russia/gdp-growth-annual#:~:text=GDP%20Annual%20Growth%20Rate%20in,the%20second%20quarter%20of%202009.">similar growth rates</a> to those seen in the West). The inclusion of Russia thus appears even more to be simply for fear-mongering effect. The inclusion of China underscores the author’s lack of understanding of decreasing marginal returns to prosperity growth. For developing countries who started out in the 1980s in the same place as the West was in the late 1940s, high growth rates are to be expected as they catch up. This pandering to fears of China and Russia feels tired and lazy. It would be so much more interesting to read even a single suggestion as to how politicians can really improve prosperity and wellbeing in the 21st century.</p><p>So here are my suggestions. If the author really wants to see declining Western powers like the UK and US return their populations to prosperity, they should re-invest in the institutions that enabled that growth in the first place, and have since withered in decay: education and healthcare chief among them. Proposals like Elizabeth Warren’s <a href="https://2020.elizabethwarren.com/toolkit/umt">2 cent wealth tax plan</a> in the US would go a long way towards reinvigorating these vital institutions. The only thing preventing such measures is greed: the top 1% in the US and UK currently have more than they ever have, and give generously to politicians who keep their taxes low. Taking even a small step to rectify this obscene imbalance is akin to stopping a bully. We must not pretend that it is theft.</p><p>Taxing top earners and corporations more, asking them to pay their fair share, would direct resources to services that not only improve the wellbeing of the populous, but also serve as the foundation for the growth-driving innovations the author nostalgically pines for. It is much harder for an entrepreneur to bring a great innovation to market while saddled with monthly school loan payments, or without the security of healthcare.</p><p>Indeed, the lack of these basic services could even be exactly what’s been inhibiting growth today. In the 60s-80s, when economic inequality was low, more people could afford to buy new things and start new businesses, spurring economic growth. Returning income and corporate tax rates to their pre-1980s levels would provide this foundation needed for a new, more sustainable increase in growth — growth not measured by GDP, but by sustainable prosperity increases. Community initiatives and care work would be valued, and we would assure prosperity of the next generation and stop overshooting our one planet. To me, that sure sounds a lot more like vitality.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e91ef10af8ea" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Data Science Expectations Gap]]></title>
            <link>https://rachelkberryman.medium.com/the-data-science-expectations-gap-227610adbc3f?source=rss-3f9e7bda6adb------2</link>
            <guid isPermaLink="false">https://medium.com/p/227610adbc3f</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[hiring]]></category>
            <category><![CDATA[interview]]></category>
            <category><![CDATA[data-scientist]]></category>
            <category><![CDATA[interview-questions]]></category>
            <dc:creator><![CDATA[Rachel Berryman]]></dc:creator>
            <pubDate>Mon, 05 Sep 2022 11:57:01 GMT</pubDate>
            <atom:updated>2022-09-05T11:57:01.251Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QK5BpzY-pO1eat9UQXPtmw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@farnoosh95?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Farnoosh Abdollahi</a> on <a href="https://unsplash.com/s/photos/gap?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>What’s causing the high divorce rate in data science: and how to cure it</p><p>The story goes something like this:</p><p>An eager young Data Scientist, fresh out of one of the new master’s programs in data science or a bootcamp, lands their first job. As a newbie in the field, they’re happy to have landed <em>any</em> job, even if it means a start-up where they’re the first data scientist (or one of a small team of 2–3). After all, they’ll finally get to apply everything they’ve learned on real data.</p><p>As their start date approaches, their excitement grows. They read new papers and practice with Kaggle challenges, making sure they’re up to date on the newest advances in transformers and neural network architectures that they’ll surely need in their new role. They know how to code, of course, but are used to doing it alone. They work with tools like Jupyter notebooks and write code that only they themselves will see. They typically have not written production software. After all, this don’t see that as their job: the title itself contains the word <em>scientist</em>, so that’s exactly what they believe they should doing: experimenting with models and trying things out.They can’t wait to start experimenting on real-life data.</p><p>Across town, their future boss is also anxiously awaiting the arrival of their new employee. They’ve read about how hard it is to find good data talent, and are so excited to have found someone who can finally help them make data-driven decisions and drive revenue growth with their AI and recommendation algorithms. They imagine their e-commerce website becoming the new Amazon, with basket sizes exploding from customers buying more products thanks to their personalized recommendations. They imagine the Data Scientist coming into their office with new insights: “Sir, I’ve thought about your strategy to open two new stores in Toronto. But the AI model suggest Montreal will be more profitable in the coming years”.</p><p>This story, full of hope and promise on both sides, almost always ends the same way (and almost always about 1 year to the day after it started). What started as a happy marriage typical ends in divorce: with both sides unhappy and blaming each other. The data scientist laments that their bosses didn’t understand the process. First of all, they didn’t even have projects to work on. <em>What kind of company hires a data scientist when they have no use cases?</em> they think to themselves. Their boss had simply pointed them at a database and left them to their own devices. Their boss didn’t understand the graphs and visualizations about train and test accuracy, and decreasing loss over the training period. They weren’t excited by the results. They often look confused and asked, “but.. when will it work?”, meaning when will the model start making predictions in real time. Yet, they hadn’t hired an ML engineer to put the model into production, and the software engineers at the company don’t want to manage ML applications. They said: “Isn’t that what a data scientist is for?”</p><p>Their boss tells a different story. They simply had the wrong guy. Although he had a PhD in statistics, he couldn’t talk to the business. He couldn’t make his work relevant. When they let him loose on the data, he complained that the data quality was bad. He said he was spending nearly all of his time doing data cleaning, and that that isn’t his job. <em>What is his job, then?</em> they wondered. What’s worse, he couldn’t <em>really</em> code. Sure, he could train models, but he wasn’t able to make them usable anywhere beyond his own machine. The company’s software engineers tried to help him, but lamented that his code was atrocious, and he’d never worked with Git. He hadn’t exhibited the proactivity the boss had expected: where were these data-driven decisions they were promised?</p><p>It would be hard to count the number of times I’ve heard this story played out in growing the data science community in Berlin. What causes it, and how can these unhappy pairings be fixed? Or at least, prevented?</p><h3>The Data Science Expectations Gap</h3><p>The two main characters in the story above had very different definitions of the term “Data Scientist”. This difference comes from their background, and how they came to learn about data science. The data scientist likely learned about data science in an academic setting: either in a master’s or bachelor’s degree, or in a bootcamp. Here, they spent most of their time fine-tuning machine learning models, with a smaller portion spent learning how to make visualizations and do statistical analysis. They likely learned where data scientists fit in the wider data ecosystem: that they get cleaned, prepared training datasets from a data engineer, and maybe even some initial insights from a data analyst. They see their role as applying and improving machine learning models: experimenting with different combinations of features, and different algorithms. After all, this is where the <em>science</em> part of data science!</p><p>However, this is a very narrow description of data science. It also represents quite a specific role: that of the Research Data Scientist (also called Research Scientist). This person typically has a PhD in a quantitative subject like statistics or applied mathematics. Their specialty is exactly that, applied research: reading research papers (and, depending on the organization, even writing them), and applying what they’ve read. They focus almost entirely n the model training and development part of the overall data science project life cycle.</p><p>This role does exist, but typically only in organizations with high AI maturity. These include AI-first companies like those working on autonomous driving, organizations with an extremely large data team and who have been deploying AI to production for years (like Zalando), and of course research-focused organizations like DeepMind and OpenAI.</p><p>This role typically does <em>not</em> exist in organizations that are not AI-first or without very high AI maturity. For the average company applying their first ML use cases, this role is not relevant. For the first 3–5 years at least, companies starting out with data science will typically only need to hire Impact Data Scientists.</p><p>Impact Data Scientists have a very different profile than Research Data Scientists. They have stronger coding skills and work well in cross-functional teams, where they work with software engineers and other profiles like user experience designers. They spend a lot of their time talking with their stakeholders and potential customers about what kinds of solutions they would like to have.</p><p>A quick note about the terminology used here: I am absolutely not saying that Research Data Scientists do not have impact in the organizations in which they work. Rather, what is meant by this term is that the <em>main goal</em> of Impact Data Scientists is to bring business impact: meaning revenue increased, or costs reduced. Bringing business value trumps any research-related goals, like trying new algorithms or developing new methods. This often means using more basic methods, or even not using machine learning at all.</p><p><strong>The data science expectation gap arises from too many Research Data Scientists entering the market, and applying for jobs that are actually Impact Data Scientist roles</strong>. On the other side, too many companies are unclear about the requirements for the data scientists they want to hire. <strong>Companies hire Research Data Scientists when they actually need Impact Data Scientists</strong>.</p><h3>How to Bridge the Gap</h3><p>Clearly, not all data science engagements are doomed to fail. There are plenty of companies who have successfully built and implemented ML solutions. They managed to overcome this expectation gap. In these cases, both sides have to be willing to compromise and have to learn to speak each other’s language.</p><h4>For Business Leaders</h4><p>Many managers hiring their first data scientists don’t understand the difference between the data science profiles mentioned above. Sometimes, they also don’t understand the difference between each of these profiles and other data roles like data engineers and analysts. They often have not clearly defined what they want their data hires to achieve. One common misunderstanding is the hiring manager expecting this data scientist to do what should actually be multiple roles: from data engineering, to analytics and predictive modelling, and finally to machine learning engineering: deploying the model in production, yet only naming the roles “data scientist” in the job description.</p><p>What’s more, the manager often expects that data scientists find their own use cases to work on. This strategy fails because the average Research Data Scientist will choose to work on projects that they find interesting: particularly those where they can apply advanced machine learning methods. These use cases are typically not the same as those that bring the most business value, especially in companies with low AI maturity. Indeed, use cases that bring the most business value are frequently “low-hanging fruits” that require more basic solutions like automation. Data scientists need support from the business to decide which uses cases would bring the most value. For this, they need data science managers who can act as translators between them and the customers they will serve. These managers can also be cross-functional product owners who are data literate and understand the basics of data science and machine learning. Without valuable use cases, the data scientist cannot bring value! It is not the job of a first data science hire to bridge this gap to the business and assess use case value.</p><p>This is why companies <em>shouldn’t</em> hire an internal data scientist right away, for the first ML use case in the company. Why not? Because for a data science use case to be successful, it needs a strong product owner or lead: one who really understands how the data science process works, and can translate between the business and the data science team. This project or product lead also needs to build experience. This is very hard to do with a brand-new data scientist. A better approach is to first partner with a data science consulting company that can support on these first use cases. This helps the product owner or manager gain experience they will need to lead their first data science hire. These initial projects with an external partner will also help the data science project manager identify issues that will block any data hires they make in the future, like data access problems, data quality issues, and workflow issues such as not having access to Python or coding environments on the company machine.</p><p>Managers must take time to educate themselves about the role of a data science manager: how to find use cases, how to support a data scientist during a project, how to advocate for them if they’re working in a cross-functional product team, and what it takes to get a machine learning use case to production. These experiences will help them better understand what specific competencies they need to look for in their first internal data science hire. Working with a consulting company on the first use cases can also help with the hiring stage, as they can assist with technical interviews and screening candidates. They can also help write targeted job advertisements that attract Impact Data Scientists.</p><p>Managers hiring Impact data scientists need to change how they conduct interviews. For these types of hires, deep mathematical knowledge of the workings of convolutional layers is not nearly important as knowing how to explain technical topics, good communication, and getting quick results. Interviews for Impact Data Scientists should include exercises that test a candidate’s ability to delight stakeholders with their communication, and to see that they start by approaching a problem with the simplest solution possible. Asking deep technical questions that focus on theory doesn’t match what this profile will do in their daily job.</p><p>Here are some sample interview questions for Managers interviewing Impact Data Scientists, with ideal responses from candidates.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gsx-olnClUGS6XLu88-FXg.png" /></figure><p>This answer shows that data scientist is laser-focused on business impact. They are not interested in applying AI for the sake of it, yet also want to help people develop their interests and nurture motivation to learn more about machine learning.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X4YKPjuOlp7lEf0iidNUMQ.png" /></figure><p>This applicant is focused on getting models into production, where they can bring value. They care about how the ML solutions they build fit into larger systems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XocYIlRl5Jr3Av9Dn6XtrA.png" /></figure><p>This applicant not only shows their focus on business value, they also show their ability to grow and take feedback.</p><h4>For Data Scientists</h4><p>Data scientists need self-awareness about what kind of data scientist they are before they can sell themselves to companies as the right fit for them. For the past 3 years, I have mentored junior data scientists entering the field after graduating from a bootcamp They often say that they apply to dozens of roles, only to get rejected. But these “applications” typically involve clicking the “EasyApply” button on LinkedIn, rather than taking the time to carefully apply to each company and to consider if that particular data science roll is really a match for their skills. They often complain that many roles require a PhD, without recognizing that these are Research Data Scientist roles, which they shouldn’t focus on if they don’t fit that profile.</p><p>Many data scientists starting out in the field think a lot about what companies should give them: clean datasets, well-defined use cases, top hardware, and above all, lots of new knowledge and experience. They then get frustrated when companies don’t deliver on these high expectations.</p><p>Data scientists would do well to approach companies with a different mindset. They should ask themselves, “how can I bring value to this company? How can I solve their most pressing challenges?”. Coming from this mindset will help data scientists align with what companies expect from them. It will also help them break out of the “tech-first” mindset. A data scientist with this mindset focuses on what technical approach they want to try, like implementing a new paper, and then tries to find a use case that fits the technology. The value-first mindset means letting go of the expectation to always use machine learning, and focuses instead on what’s needed to bring value.</p><h4>Bridging the Gap</h4><p>Both data scientists and managers need to understand the difference between Impact Data Scientist and Research Data Scientist roles. Hiring managers and companies starting out with ML should first work with more experienced consultants and freelancers before hiring internally. Data scientists should know which of these two roles they fit, and apply only to roles that fit their profile.</p><p>With these small tweaks, both sides come much closer to speaking the same language and understanding each other, leading to more fruitful engagements for all.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=227610adbc3f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What’s happening in German politics right now, explained for Americans]]></title>
            <link>https://rachelkberryman.medium.com/whats-happening-in-german-politics-right-now-explained-for-americans-1221b947790e?source=rss-3f9e7bda6adb------2</link>
            <guid isPermaLink="false">https://medium.com/p/1221b947790e</guid>
            <category><![CDATA[germany]]></category>
            <category><![CDATA[political-science]]></category>
            <category><![CDATA[usa]]></category>
            <category><![CDATA[politics]]></category>
            <category><![CDATA[europe]]></category>
            <dc:creator><![CDATA[Rachel Berryman]]></dc:creator>
            <pubDate>Tue, 25 May 2021 13:21:15 GMT</pubDate>
            <atom:updated>2021-05-25T18:40:58.452Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*Qkf_JVfddf6HeexONYZPaQ.jpeg" /><figcaption>A Green Wave is coming</figcaption></figure><p>Germany is about to get a new chancellor for the first time in 16 years. Why this change could mean major disruption to the German political status quo.</p><p>When I first moved to Germany from New York in 2017, the country was in the midst of <a href="https://en.wikipedia.org/wiki/2017_German_federal_election">federal elections</a>. I spoke to friends and colleagues who were horrified at the rise of the right-wing AfD, Alternative for Germany (Alternative für Deutschland) party, decrying their supporters as facist imbeciles with no care for democracy or for their fellow man. Given the sudden shift to the right that had happened in the American election just one year earlier, I recognized their response as the same one I had seen in friends back home. They had similar fears and the same visceral sense of their country’s moral fabric being stripped away. Still, it felt different than 2016 America. The AfD, though they gained a share of seats in parliament for the first time in 2017, did not become the ruling party. Their leader did not become the head of the German government. In comparison to the drastic regime switch in the United States, the German elections of 2017 felt like a ripple on the water of an otherwise calm and predictable sea, that sea being the vast and seemingly unending reign of the CDU (Christian Democratic Union) and its sister party, the CSU (Christian Social Union), with their steadfast leader, Angela Merkel.</p><p>In the time since 2017, and particularly in the past year, this once calm and steady sea of CDU power has become increasingly turbulent. Now, as the country prepares for federal elections again this fall, the threat that the CDU/CSU’s16-year reign will end is very real — its grip on power threatened not from the far-right, but rather from tree-huggers and humanitarians: the Greens.</p><p>Starting with a brief explanation of German politics, we will look at the forces driving this seismic shift in German politics today, and what it could mean for German/American relations in the future.</p><h3>Germany’s Political System</h3><p>To understand how German politics are changing, we first have to understand how the political system works and how it differs from the American system. Like the US, Germany is a federal democracy, meaning it is made up of various states that have their own power and ability to enact laws at the state level. There are 16 German states, each with a leader (called Ministerpräsident), similar to a governor, and its own state legislature, like the US states have. At the federal level, Germany has a bicameral (2-chamber) legislature, also like the US, with a lower chamber similar to the House called the Bundestag where members serve 4-year terms, and an upper chamber called the Bundesrat. Only members of the Bundestag are elected directly; members of the Bundesrat are chosen by each individual state from their ruling parties. Thus, they do not serve fixed terms, and rather serve at the discretion of state leadership.</p><p>The head of government, called the Chancellor (Kanzler:in) is also not directly elected. Rather, she or he is elected by the Bundestag. In practice, this means that the head of the party that received the most seats in the Budestag will become Chancellor.</p><p>For the past 16 years, that party has been the CDU/CSU. You might be asking yourself now why the distinction exists, and what exactly the difference is between the CDU and the CSU. The CDU is the center-right party in Germany, and is active in all states <em>except</em> Bavaria. Bavaria has its own center-right party, which is the CSU. To understand how this works, imagine if the Democratic Party didn’t exist in Texas, and Texas had its own party called the Texas Democrats (given the state’s penchant for <a href="https://en.wikipedia.org/wiki/Texas_secession_movements#:~:text=Texas%20secession%20movements%20refer%20to,sovereign%20state%20since%20the%201990s.">separatism</a>, <a href="https://www.texastribune.org/2011/02/08/texplainer-why-does-texas-have-its-own-power-grid/">isolated energy grid</a>, and <a href="https://www.reddit.com/r/houston/comments/42tqqp/honest_question_why_are_people_in_texas_so/">obsession with flying its own flag</a>, this isn’t that difficult to imagine). What would a Texas Democratic Party look like? Well, it would likely be very pro-Texas, and thus a bit more conservative on issues like states rights. However, it would also be unlikely to put forward its own candidate on the ballot for presidential elections — voters in Minnesota or New York would be unlikely to think that a candidate running as a Texas Democrat, rather than a “regular” Democrat, would fight for their best interest. Thus, the Texas Democrats would likely support the national Democrats in the presidential election, while still sending only Texas Democrats to state-wide offices like the Senate and House, where they would vote (mostly) in line with the “regular” Democrats to advance policy goals. This is roughly how the relationship between the CDU and CSU has tended to work since the end of the Second World War.</p><p>Although the CDU/CSU alliance has dominated German politics for nearly the past 2 decades, that certainly doesn’t mean that Germany is a one-party state. The CDU hasn’t had enough seats to govern on their own in quite some years. In order to pass laws, they have formed coalitions with other parties, namely the center-left SPD (Social Democratic Party). Other challenger parties have been slowly gaining seats, like the AfD in 2017, as well as the far-left party Die Linke (which literally translates to “The Left”), and the pro-business FDP (Free Democrats).</p><p>Another point worth mentioning about the German political system is that state and local elections happen on different days throughout the year, depending on the state. This will become important later one when we discuss what’s happened so far this year.</p><h3>Current Political Sentiment and Signs of Change</h3><p>It’s impossible to discuss the current changes in public opinion regarding politics in Germany without talking about the Covid-19 pandemic. The pandemic altered nearly every facet of our daily lives. It also hastened many trends that were already happening, like those towards remote work and studies and the popularity of travel by bicycle.</p><p>These societal trends seem likely to have had a hand in hastening a political trend that had already been brewing: that of the decline of the traditional center parties, and rise newcomer parties.</p><p>As in many other countries, like the US, the pandemic has shown fissures in the status quo of German social and political life. As in America, <a href="https://www.tagesschau.de/inland/corona-frauen-101.html">women took on more</a> of the additional domestic duties brought on by closed schools and daycares. An underpaid and short-staffed system of care workers buckled under the weight of the additional burden, particularly in elderly care facilities. Many people in society began to ask the same questions being asked in societies around the world: Is there not a better way? Can the post-pandemic future not look different — look better — than the status quo from “BC” (Before Corona)?</p><p>As they were asking these questions, they looked to their country’s leadership for answers. What they found did not placate them, and rather gave them new reasons to worry. The CDU-led coalition government continually flip-flopped on Covid mitigation policy, often ignoring the latest scientific advances in favor of outdated methods. And then came the scandal.</p><p>Multiple members of the CDU were accused of profiting, to the tune of millions of euros, from kickbacks from companies they helped broker deals with to supply the country with masks. Multiple politicians from the party were accused of profiting, leading the public to <a href="https://www.dw.com/en/opinion-masks-money-and-a-lack-of-morals-among-germanys-conservatives/a-56820674">lose trust</a> in the party and making the scandal seem more widespread and less like a one-off issue of poor judgement by one or two politicians.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/828/1*Hv-wuUjxIemx6KyQAK2ybQ.png" /><figcaption>Meme by @thefunnyauslander on Instagram making fun of the mask scandal. It says “Germany: We need faster vaccinations and rapid tests. CDU/CSU: Best I can do is Mask Scandal”</figcaption></figure><p>If the fallout from the mask scandal were all it had to overcome, the CDU/CSU probably wouldn’t have lost so much of the public’s trust. Unfortunately that wasn’t the case, especially not concerning the topic du jour: the second wave of the Covid-19 pandemic. Germany entered a strict lockdown in November, partially because of rising case numbers, and partially over fears of an “explosion” of cases from the B.1.1.7 “British” variant. Months into the lockdown, the public grew angry as the CDU-led government repeatedly backtracked, changed strategy, and pushed the yardstick for re-opening further and further away. The 7-day incidence rate needed to re-open outdoor dining, for example, bounced from 50 cases per 100,000 residents, to 35, then to 100. The government also fumbled its nationwide testing rollout, and, like the rest of the EU, its vaccine campaign.</p><p>At the same time, the party was split by bitter in-fighting over who would become the next Chancellor candidate. The CDU narrowly voted for Arnie Laschet over Friedrich Merz, a more right-leaning figure with a history of <a href="https://de.wikipedia.org/wiki/Friedrich_Merz#Politische_Positionen,_Kontroversen_und_Kritik">gaffes</a> that would rival those of <a href="https://www.abc.net.au/news/2021-04-11/prince-philip-gaffes-complicate-his-legacy/7993970#:~:text=Prince%20Philip%2C%20who%20died%20on,which%20he%20himself%20was%20aware.">Prince Phillip</a>. Even after Laschet was chosen to lead the CDU, the “Chancellor question” wasn’t answered: the leader of the CSU, Markus Söder, expressed his interest in leading the sister-party partnership, citing higher poll numbers supporting him over Laschet. But remember the “Texas Democrats” example I gave before — although he was generally popular at the time, few in the CDU/CSU parties felt that Söder could succeed on a national scale. Laschet was finally chosen as the CDU/CSU candidate following this very long and very public display of party disunity.</p><p>Still, Laschet remains unpopular as an individual, and the CDU/CSU remain unpopular as a party. So if people are losing interest in the CDU, the logical question is, to whom are they turning?</p><h3>The Rise of the Greens</h3><p>The answer is: they’re going Green. The Green party has been steadily gaining popularity in Germany since their (official) founding in 1993, fueled by an electorate that is becoming <a href="https://www.umweltbundesamt.de/themen/nachhaltigkeit-strategien-internationales/gesellschaft-erfolgreich-veraendern/umweltbewusstsein-in-deutschland">increasingly climate-conscious</a>. This year, there seems to be an especially strong interest not just in the party, but also in their party co-leader and Chancellor candidate, Annalena Baerbock. She consistently polls higher than Laschet in polls that pit them against oneanother, and against Olaf Scholz, the candidate from the SPD party, which previously provided many chancellors but have lost considerable share in the past few years.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*SIg1R7uDKy4-sxZGnL78nQ.jpeg" /><figcaption>Poll results to the question, “If you could vote for the next Chancellor directly, who would you vote for?” show the Green party’s Annalena Baerbock in the lead. 30% of respondents answered with “don’t know” (“weiß nicht”), possibly showing a sizable share of undecided voters left for the various parties to sway.</figcaption></figure><p>This has a lot of people talking, as it would be historic for many reasons. One reason is Baerbock’s age: at only 40 years old, she would be Germany’s youngest Chancellor.</p><p>A potential Green Chancellorship is also historic as it would be a first in Germany. Since the Second World War, every Chancellor (with the exception of Walter Scheel, who we won’t count since he ruled for a whopping <a href="https://de.wikipedia.org/wiki/Bundeskanzler_(Deutschland)#:~:text=Amtierende%20Bundeskanzlerin%20ist%20seit%20dem,aus%20CDU%2FCSU%20und%20SPD.">9 days</a>…) has belonged to one of two parties: the CDU (Laschet’s party) and the SPD (Scholz’s party). A Green Chancellor would thus represent a stark departure from the status quo of the last half-century. This shift can even be seen in the first pages of the party’s <a href="https://www.gruene.de/artikel/wahlprogramm-zur-bundestagswahl-2021">platform</a> (Wahlprogramm): the entire first chapter focuses on ecological goals, like transitioning away from fossil fuels more quickly and making it easy to live a climate-neutral life. Germany would be the first G7 nation to make curbing climate change its top national priority.</p><h3>What would a Green Chancellorship mean for US-German relations?</h3><p>America has long had a steady friend in Germany. Since the Second World War, US-German ties have been fostered by both parties, in both countries (the Republicans and Democrats in the US, and the SPD and CDU in Germany). This has remained true even as the Republican party in the US has moved further to the right and been openly more critical of Germany. Angela Merkel famously kept her cool when dealing with former President Trump, even after he <a href="https://www.theguardian.com/us-news/2020/jul/01/donald-trump-theresa-may-angela-merkel-setbacks">called her</a> a “loser”. However, it would be unlikely that a Green Chancellor like Baerbock would be so willing to tiptoe around and avoid confrontation with a future conservative US leader — if the German electorate votes Green this year, they will be voting for change. In the spirit of the loud-and-proud <a href="https://www.rbb24.de/politik/beitrag/2021/05/fridays-for-future-weltweiter-klimastreik-menschenkette-kanzleramt-berlin.html">Fridays for Future demonstrations</a> last year that helped advance the environmentalist agenda in Germany, they would likely expect their new leader to be unapologetically and unequivocally pro-planet. This could also lead to tensions even with the Biden administration, if they are seen to not be doing enough to stop climate change.</p><p>However, Baerbock’s past could reveal a soft spot for America and our Anglophone brethren across the pond: she <a href="https://de.wikipedia.org/wiki/Annalena_Baerbock">lived in Florida</a> for a year as a high school exchange student, and also has a master’s degree from the London School of Economics.</p><h3>How likely is a Green Chancellorship really?</h3><p><a href="https://www.tagesschau.de/inland/deutschlandtrend/deutschlandtrend-2609.html">Polls</a> currently show the Greens leading the CDU by about 3 percentage points.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*fyB7_Th1TZbiz6TDRpunYw.png" /><figcaption>Poll showing which party responders would vote for in the Federal election in September. The Chancellor candidate from the party with the most votes will become the country’s chancellor.</figcaption></figure><p>The Greens are also helped by Baerbock’s personal popularity: another poll showed that she is the second most popular politician in Germany, after only Angela Merkel.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*X7pkabWma05AwtFqSpkfNg.png" /><figcaption>Poll showing percentage of respondents who said they are satisfied with the given politician. The color of the bars represent the color of the party to which the politician belongs.</figcaption></figure><p>Even more evidence than these glowing polls exists in support of the Greens. Remember that state parliament and local government elections happen at different times of the year, depending on the state. In a Federal election year like this one, the state elections that happen earlier in the year are often seen as a barometer for checking how the public feels going into the Federal election. So far, 3 states have held elections: State parliamentary elections were held in Rheinland-Pfalz, and Baden-Württemberg, and local elections were held in Hessen. In all three states, the Greens gained significant ground over the incumbent parties.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*vq1Fh2m3YyBM6d3P6Cvs0Q.jpeg" /><figcaption>Greens gained 4 percentage points more of the vote than they had in the last election in Rheinland-Pfalz. Source: <a href="https://www.tagesschau.de/wahl/archiv/2021-03-14-LT-DE-RP/index.shtml">https://www.tagesschau.de/wahl/archiv/2021-03-14-LT-DE-RP/index.shtml</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/905/1*Q_f7USrb1wPOnCARQw9DPA.jpeg" /><figcaption>The Greens gained 2.3 percentage points over their previous vote share in Baden-Württemberg. Source: <a href="https://www.tagesschau.de/wahlarchiv/laenderparlamente/baden-wuerttemberg/">https://www.tagesschau.de/wahlarchiv/laenderparlamente/baden-wuerttemberg/</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/905/1*D0QnUEenT5XyIZ4HlIkD6w.jpeg" /><figcaption>They remained the strongest party in Baden-Württemberg. This graph shows the total share of votes in the state election, with the Greens having the strongest share. Source: <a href="https://www.tagesschau.de/wahlarchiv/laenderparlamente/baden-wuerttemberg/">https://www.tagesschau.de/wahlarchiv/laenderparlamente/baden-wuerttemberg/</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/707/1*RzyF1qPCBp9THNOI6sKG-Q.jpeg" /><figcaption>In the local elections in Hessen, the Greens made huge gains of 7 percentage points over the last election in 2016. Source: <a href="https://wahlen.statistik.hessen.de/">https://wahlen.statistik.hessen.de/</a></figcaption></figure><h3>What happened to the AfD?</h3><p>As mentioned at the start of this article, there was significant fear of the rise of the right-wing AfD party after the last federal election in 2017. There were also fears of a possible further strengthening of the AfD following the rise of the so-called “<a href="https://de.wikipedia.org/wiki/Proteste_gegen_Schutzma%C3%9Fnahmen_wegen_der_COVID-19-Pandemie_in_Deutschland">Querdenker</a>” movement. Members of the movement think of themselves as outside of mainstream politics and are particularly <a href="https://querdenken-711.de/">against state-sponsored measures</a> put in place to prevent the spread of Covid, hosting multiple rallies and protests against mask mandates and social distancing protocols. Many members of the movement, which grew in strength over the course of the pandemic, were aligned with the AfD.</p><p>However, despite these fears of increasing interest in the AfD, the first 3 elections from this year have only shown significant losses for the party. Where are all of these former AfD voters and “Querdenkers” going? The answer is, nowhere. The vast majority of voters who voted for the AfD in the previous election in both Baden-Württemberg and Rheinland-Pfalz simply fell out of the political process altogether, deciding not to vote for any party this year.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*kVFM8rqm8v_mA5EJwzUcaw.jpeg" /><figcaption>Where voters of the AfD went to since the last election in the state of Rheinland-Pfalz. They did not gain votes from any party, and a vast majority of their former voters became nonvoters (“Nichtwähler). Source: <a href="https://www.tagesschau.de/multimedia/bilder/uvotealbum-1241.html">https://www.tagesschau.de/multimedia/bilder/uvotealbum-1241.html</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*aAhRkGGMvV3P9vGZai4dCw.jpeg" /><figcaption>The same pattern can be seen in Baden-Württemberg, where over 100,000 former AfD voters stopped voting this year. Source: <a href="https://www.tagesschau.de/multimedia/bilder/uvotealbum-1239.html">https://www.tagesschau.de/multimedia/bilder/uvotealbum-1239.html</a></figcaption></figure><h3>Looking Forward</h3><p>The question remains of whether the AfD will be able to win back their former voters who fell out of the political system. At the moment, it seems that excitement is a major factor influencing German politics: the Greens have a lot of excitement, which so far does seem to be paying off, in that this excitement is also translating to votes. Both the AfD and CDU/CSU seem to be unable at the moment to elicit the excitement and enthusiasm needed to drive people to the polls for them.</p><p>Only time will tell if the Greens can use this excitement to pull of an upset and unseat the reigning CDU. Even if they don’t take the Chancellory, the change in national policy focus they set off will not go away quietly. If not fully taken over by a Green wave, the steady sea of German politics will certainly take on a greener tinge.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1221b947790e" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>