<?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[Epishow Behind the Scenes - Medium]]></title>
        <description><![CDATA[Find out what we’re building. - Medium]]></description>
        <link>https://medium.com/epishow-behind-the-scenes?source=rss----831fbd702104---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Epishow Behind the Scenes - Medium</title>
            <link>https://medium.com/epishow-behind-the-scenes?source=rss----831fbd702104---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 06:40:44 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/epishow-behind-the-scenes" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The art of the user save]]></title>
            <link>https://medium.com/epishow-behind-the-scenes/the-art-of-the-user-save-9a11546c8955?source=rss----831fbd702104---4</link>
            <guid isPermaLink="false">https://medium.com/p/9a11546c8955</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Randall Bennett]]></dc:creator>
            <pubDate>Wed, 10 Feb 2016 17:14:21 GMT</pubDate>
            <atom:updated>2016-02-10T18:06:42.601Z</atom:updated>
            <content:encoded><![CDATA[<p>One thing I’ve consciously strived to do with <a href="http://www.epishow.com/">Epishow</a> is pay really really close attention to users using the site. Not with automated tools but with human focus.</p><p>We’re small enough in scale (few dozen users?) that I can personally pay attention to anyone who makes it past the homepage. While we’re in the early days, using analytics tools to gather insights is actually really unhelpful because we don’t care about an aggregate number of users who’ve done specific actions, we care about individual users: Did Tim make it all the way through to start an Epishow? Why or why not?</p><p>The answer, unfortunately, is often “because things are broken.” I’m not a school-trained coder, and a lot of the code I’ve written is well intentioned but buggy. As a result, some stuff is just flat broke.</p><p>We track a user’s progress by simply notifying a slack channel (we call it #fyi) whenever a user does an event we care about. I think the right approach is to probably track it in two or three places: Slack, Heap and Intercom. That’s probably the holy trinity of micro-user-research.</p><p>Once you’ve figured out what’s broken, it’s time to reach out. I learned this trick with Vidpresso’s early days. Even if your software is objectively horrible, if it solves one specific need and does that really well, people will put up with it to a point. The interesting thing is you can make up for all the parts that suck by being extra attentive.</p><p>In our case, if someone makes it to the broadcast page but doesn’t make it all the way through, I try to reach out. First, if they’ve logged in with Twitter, I start there. Second, I shoot them an email. “Hey saw you tried Epishow… looked like you might have had problems. Did it work ok?”</p><p>The answer back is usually “no, it had problem x.” From there I immediately diagnose and fix problem x for them. This kind of customer service feels insanely foreign. Imagine if Comcast had the kind of care you just showed. You’d love them! They’ll stick with you in the future in spite of yourself.</p><p>We still do this with Vidpresso. We have a product that was being installed at a station which used software we hadn’t integrated in production before. It wasn’t going well, so we hopped in our car and drove there. They were surprised and excited because we did it without asking first. Then we fixed the problem. That kind of customer service is pretty rare and as a result they gave us more benefit of the doubt.</p><p>Try it! It doesn’t scale at all, but when you have so many requests you can’t do anything else, you can either automate (consumer, low $ biz) or hire people (SaaS monthly recurring) to replace you.</p><p>You probably won’t have that problem.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9a11546c8955" width="1" height="1" alt=""><hr><p><a href="https://medium.com/epishow-behind-the-scenes/the-art-of-the-user-save-9a11546c8955">The art of the user save</a> was originally published in <a href="https://medium.com/epishow-behind-the-scenes">Epishow Behind the Scenes</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Podcasting means serving files, and that’s missing the point.]]></title>
            <link>https://medium.com/epishow-behind-the-scenes/podcasting-means-serving-files-and-that-s-missing-the-point-4a94e22e1e4c?source=rss----831fbd702104---4</link>
            <guid isPermaLink="false">https://medium.com/p/4a94e22e1e4c</guid>
            <category><![CDATA[podcast]]></category>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Randall Bennett]]></dc:creator>
            <pubDate>Tue, 13 Oct 2015 20:06:51 GMT</pubDate>
            <atom:updated>2015-10-13T20:06:51.343Z</atom:updated>
            <content:encoded><![CDATA[<p>Originally published: Aug 21, 2015</p><p>I’ve seen a bunch of podcast services that new podcasters are trying to use to create, distribute and monetize their podcasts.</p><p>Among them:</p><ul><li><a href="https://www.libsyn.com/">libsyn</a></li><li><a href="https://www.blubrry.com/">Blubrry</a></li><li><a href="https://www.spreaker.com/">Spreaker</a></li><li><a href="http://zencastr.com/">Zencastr</a></li></ul><p>I’ll keep going as I find more to add to the list, but here’s the fundamental problem: You’re all making files and making files is not what ‘web audio’ or ‘web video’ need.</p><p>Podcasting today is seriously like “web zines” or whatever they were called in 1995. It’s like someone creating a static file, uploading it to a server, then counting the number of times that file was downloaded. And people are complaining about advertisers being reticent to work with podcasters because the metrics are unreliable or scary.</p><p>Seriously, it feels just like 1995. We’re crafting files by hand, uploading them somewhere, and then hoping they get served.</p><p>What changed? I think early CMSes like PHPnuke or whatever that became state of the art around the time <a href="https://en.wikipedia.org/wiki/Viaweb">PG was messing around with Viaweb</a>. The important part about Viaweb and other early CGI or perl based CMSes is it transitioned people from creating files of HTML, uploading them to a server, etc., into creating data. The difference is most starkly contrasted between GeoCities and Wordpress. GeoCities sites were “personal websites” often times, but lacked any sort of structure. And updating them required at least a beginners level of knowledge of HTML. And those HTML pages were copied / pasted into each individual page… no pages were truly related except that they had links to each other. Each page was generated by hand.</p><p>This is what audio and video production are like today. Specifically with audio podcasts… we’re hamstrung to this insane standard that was haphazardly created by Apple. Odeo and other companies were smart to get out of podcasting, because guess what, they were at least 10 years too early. The big technology triggers that are making our dream of Audio and Video as client assembled thing a reality:</p><ul><li>Node.js and technologies like it (go, rust, streamy async by default stuff)</li><li>Kurento</li></ul><p>I live my life by the <a href="http://www.brepettis.com/blog/2009/3/3/the-cult-of-done-manifesto.html">cult of done manifesto</a> so I don’t really have a ton of time to edit this, but I’m definitely going to do an audio show about it today with Justin. Link will be here.</p><p>Node (and other async pipelining technologies) are huge because they let us do things faster and the way we need… as needed. With node we can start encoding an MP3 and pipeline the result to the user as its being served (mostly). Kurento is the capture side… it’s relatively simplistic to record one or two people, and with time and resources we’ll figure out efficient ways to record even more. And livestream the whole thing. And enable audience participation.</p><p>The thing to focus in on: Audio and Video need to be data descriptions of what the contents of the presentation are. Audio needs to be a series of clips describing in and out times, with potentially applicable filters and other presentation descriptors.</p><p>Pauli from our team is extremely close to releasing the first version of our new audio server, which we’re calling audioserv. It’s going to handle dynamically assembling (and caching, natch) audio shows with multiple clips and different audio levels. Kurento is handling the capture of audio, then our audioserv is processing and storing those “raw” files, and we’re mixing them as needed on-the-fly.</p><p>There’s a huge chance this strategy is going to fail, or not scale, or something, but it’s also going to do something that to my knowledge has only been done a few times: Dynamically assemble clips together. Previous to this, Castfire, a company at which I previously worked, did this for audio and video, but they didn’t handle overlapping clips. That’s a big part. From this we’ll also be able to generate high-quality “highlight reels”, or even create a pandora for audio of sorts, where it’s one huge stream of shorter-than-the-whole-podcast clips.</p><p>It’s too early to really get this stuff right. We’re going to screw it up. But at least we’re trying. I wish more people would think about what’s possible instead of what is. Today audio is hard to create, and very few people feel like they can do it because they feel limited by their tools. We’re hopefully going to do for audio what Wordpress, Twitter and Blogger did for textual internet content.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4a94e22e1e4c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/epishow-behind-the-scenes/podcasting-means-serving-files-and-that-s-missing-the-point-4a94e22e1e4c">Podcasting means serving files, and that’s missing the point.</a> was originally published in <a href="https://medium.com/epishow-behind-the-scenes">Epishow Behind the Scenes</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>