Twitter, Mastodon and the Fediverse: I Have Questions

Just under sixteen years ago, I joined Twitter. I did it reluctantly, because initially it seemed both trivial and self-important. Which was both right and wrong, as it turned out. Since that time, a lot has happened. The business I helped found has grown – thanks in part to Twitter. I met and married a girl – not thanks to Twitter, but also not in spite of it. And we had a child whose exploits and repeated ownings of me are objectively speaking the most popular things I’ve posted to that service.

I’ve got some history with Twitter, is what I’m saying. It’s why like a lot of people I’m sad about how completely, appallingly and predictably unqualified the new owner is to run it, and how shocked I am in spite of all of that that he’s driven a service widely regarded as a global town square to the brink of ruin less than three weeks after acquiring it.

If you’ve read any history, you’ve come across accounts of people wandering through abandoned cities. Once thriving population centers, they were where thousands, tens of thousands or millions lived out their lives. Fast forward some period of time – in some cases a very brief period of time – and they were ghost towns, emptied of people, trade and history. The most interesting and unanswerable question wasn’t when the city had been abandoned, or even why. The question was who was the last to leave.

No one necessarily expected to be asking that question about Twitter in November of 2022, at least until it became clear that the eventual acquirer had pushed his trolling too far and would be forced by courts to complete the acquisition lest even more embarrassing texts come to light, but here we are.

I’ll save the elegy for Twitter for another time. It’s not dead yet, after all, and it might feel like going for a walk. Like others, I can’t see it coming back from all of this, but it certainly wouldn’t be the first time one of my predictions missed the mark.

Instead I want to talk briefly about the primary proposed alternative, at least in my circles: Mastodon, and the distributed network it powers, the Fediverse. I’m not going to write a primer on that – numerous outlets have already done that, e.g. Wired – but I do have questions. And given that some friends are scheduled to chat about all of this tomorrow at noon ET, I thought I’d throw some of these questions out there for discussion. I’d ask them myself, but given that said discussion is smack in the middle of a consult I’m not able to attend.

As I consider the Fediverse, then, and Mastodon, I’m less focused on the experience and what it’s like for me personally than I am occupied with questions. Questions I’m trying to answer to understand what the future of both might be, and therefore whether and how to invest my time accordingly. Here are a few of those.

How accessible is the Fediverse?

If you’re a technical user looking to communicate largely with other technical followers, this may not be an issue. The technically savvy can probably traverse the gap between the world in which you just sign up for Twitter and the one in which you have to pick an individual server from one of dozens.

But speaking as someone who follows a lot of different communities on Twitter ranging from my technologist peers to baseball writers to shark researchers to meteorologists to historians to national security professionals, it’s not clear if all or even most of these populations will be able to make this jump seamlessly, if ever.

The difference between “go signup at” and “here are a choice of different servers with different communities and rules” might not seem insurmountable, but it’s certainly not ideal. So I’m curious as to how the on ramps will be bridged over time.

What are the impacts and costs of a federated network?

While no expert on the Fediverse, in spite of having been a nominal member of it for five years or so, a couple of things have jumped out about the network during my usage.

  1. There are UI costs to a federated, decentralized network. In some cases, I can click a button and follow a user. In other cases, I have to cut and paste an identity URL into a search box, and click through twice to follow a user. In still other cases, I have to do the above, but first enter a URL of my own to follow a given user. None of these, again, are unsolvable UI problems. But they are highly likely to be discouraging to the casual user, particularly those accustomed to centralized services like Facebook or Twitter.
  2. One of the founding principles of the Fediverse is that each server can have its own culture, norms and expectations. Centralization is no panacea when it comes to content moderation, of course; Facebook, Twitter and the like are regularly referred to as cesspools, and with good cause.

    The Fediverse’s reaction to this, apparently, has been to allow individual servers to determine their own policies on content moderation. Which seems like an improvement, until it is more properly revealed as a tradeoff. In one example that is preventing people I know from joining the Fediverse – policies intended to limit the impact of racist discussion and behaviors as a means of protecting sensitive users from harm in practice have meant that People of Color (POC) are prohibited from speaking freely about their own lived experiences with racism, because non-POC don’t want to have to see it. That’s not the intent, but the intent doesn’t matter. It’s bad.

    More subtly, there are policies like this one from

    “Do not “shitpost” – while humorous posts are allowed, and actually encouraged, there is no place for “shitposting” on Fosstodon.”

    On the one hand, as someone who doesn’t particularly enjoy shitposting, this sounds good on paper. On the other hand, it begs the question: who is determining what is and is not a “shitpost?” Another question: who has the time to evaluate all of the posts on a given network to determine whether it’s shitposting, and are all of the accounts likely to be evaluated similarly?

    In many respects, the idea of federated cultural policies reminds me of the situation with alcohol in the United States. Because alcohol is regulated not at the federal but the state level in the US, the laws vary. Some states allow spirits of any kind to be shipped. Other states allow wine, but no other spirits. Other states prohibit anything from being shipped. I went to college in a state that did not allow alcohol to be sold on Sunday, so instead we drove across the border to a state next door that did. Some states have a maximum on the alcohol content by volume of a given beer; per the Sam Adams website, for example, their Utopia beer is illegal in Alabama, Arkansas, Georgia, Idaho, Missouri, Mississippi, Montana, North Carolina, New Hampshire, Oklahoma, Oregon, South Carolina, Utah, Vermont and West Virginia.

    Regardless of whether one drinks alcohol (I do) or Utopias (I don’t), the mishmash of conflicting rules and regulations seems both arbitrary and problematic for everyone involved to navigate.

    But in a federated world, this is the reality.
  3. Processing the deltas in expectations between federated servers might be challenging as an individual. What further complicates matters is the fact the behaviors of others on your server may be held against you. If someone else on Twitter behaves badly, other than a general sense that it makes Twitter a less pleasant place to be overall, the impact on my account is zero. I learned this week, however, that because the instance I joined five years ago lacks content moderation resources (we’ll come back to that shortly), my posts may be de-emphasized or even outright banned from other servers

    The good news is that the Fediverse makes it possible to up and move from one server to another with no loss of followers. But while I might be willing to undertake that given the right incentives, how many ordinary, less dedicated social media users would be?

    And given that servers are largely community run and thus – as pictured above – likely to be short on moderators, how long will it be before getting particular servers de-federated becomes weaponizable – with bad actors descending in concert on a targeted server with the express goal of having it become uninhabitable and defederated?

What about verification?

One of the dumbest decisions of the new Twitter owner’s tenure has been to mess around with account verification – the inevitable result of which has been chaos. See, for example, a random account temporarily shaving $15 billion off Eli Lilly’s market cap with a single tweet.

This is a theoretically solvable problem in a centralized universe such as Twitter – at least if you’re smart enough to not light it on fire because you haven’t thought things through. How this would or could be solved in the Fediverse is unclear, however. Registering a household brand name on any given server is a trivial thing, as is registering that brand name on lots of them.

How should users determine which identities they can trust? Should that burden be on them in the first place? Or is the idea that the Fediverse isn’t a network for brands at all – in which case, how do I actually get somebody like Comcast or Spectrum to respond to a support request? Because while Twitter’s good for that at present email sure as hell doesn’t work.

What are the economics of the Fediverse?

I remember talking to a friend working in the investment space many years ago about open source software. He could not comprehend or even believe in a world in which groups of software developers, often working without compensation on a volunteer basis, could possibly create something that would outperform a competitive project produced by a commercial entity. While I understood his skepticism, that model did not then and never has confused me. Obviously the monetary investments into open source software are and have been foundational to its ascent, but the intellectual challenge of solving interesting problems is something that has and presumably always will attract the interest of software developers, whether they’re paid for the work or not.

In my experience, however, this is not the case for the more mundane, and often thankless, task of operating that software on an ongoing basis. After the novelty of spinning something up wears off, it becomes tiring to run and maintain software. Which is why in most scenarios in which software needs to be relied upon, someone is paid to do the job of running it.

As a side note, if you’re reading this and objecting – “but I run my own email server!” – that’s great and I respect your dedication while questioning your priorities, but it is my sad duty to inform you are the exception that proves the rule.

Regardless, this is why I’ve been curious about the Fediverse and Mastodon’s financial footing. The software development, as far as I can tell, has been mostly crowd sourced. Which, ok. Maybe that can be made to work. But who is running all of the federated Mastodon servers? Will they keep running them? Why will they keep running them? What if they don’t keep running them?

Also, besides the cost of the time spent keeping the software up and running, there is the expense of hosting. I haven’t run a Mastodon server myself, but every indication I have seen to date by those that have suggests that it is likely to be non-trivial in cost for anything sizable.

For most social media historically, the cost of the service has been indirectly born by advertisers. Ad-based models are self-evidently problematic for any number of well known reasons. But user funded services come with their own set of tradeoffs, most obviously by privileging those able to pay for a non-critical technology service ultimately resulting in a less diverse user base.

I understand, for better and for worse, the business model of ad-based social networks. What is not clear to me, at least at present, are the economics of the Fediverse. And I’m not the only one with questions on that subject:

And what about funding?

Andrew Couts [WIRED senior editor of security]: I don’t actually know the answer to that. It’s a nonprofit, so I believe it’s mostly crowdsourced funding. I know they have a Patreon page, and so that’s who you would be giving to. You would be giving to the main Mastodon nonprofit. But besides that, I’m sure that there are Patreon pages for individual servers, individual instances, and it’s mostly just a crowdfunded thing. Nobody owns it, so there’s nobody to pay or anything of that nature. You’re not going to be charged $8 for using Mastodon. And if you were, you could move to another Mastodon server and ignore that.

Even if you wanted to monetize Mastodon in traditional ways, its federated nature might act to limit the ability to generate revenue. Much as the asinine decentralized pools of subsets of the population that represent regional health care networks here in the United States limit their respective ability to negotiate with suppliers and in so doing drive up costs, so too do the smaller pools of federated users constrain the economic potential of any given Fediverse server versus the centralized populations of Facebook, Twitter, et al.

Which might sound perfectly acceptable and even attractive if you don’t want to see ads ever, but then you need to figure out who’s going to pay for everything if it’s not advertisers. Because even if the software development costs can be figured out, the hosting and operational costs – not to mention things like content moderation, trust and safety, security and so on, which are both hard and expensive – have no clear solution from where I’m sitting.

So even if I want to believe, and I do want to believe, my fundamental question is: how is all of this going to work economically?

Last question: “Toots?”

Really? Toots? Toots? And I thought “tweets” was silly.

All of the above said, if Mastodon could import my Twitter archive, I’d definitely do that because skeptical as I am, I’m rooting for it – “toots” notwithstanding.

Our New Network Solution: An Amplifi Review

IMG_20180117_101733 (1)

Back in early January, the Nest camera that serves as our baby monitor went offline in the middle of the night. By itself, this wouldn’t be noteworthy, but the problem was that this had become routine. And any routine that involves an already sleep deprived parent having to wake up in the middle of the night and mess around with network connections isn’t going to be a routine for long. Devices from Nest units to our laptops would get hung up or knocked off our network entirely, several times a day every day.

Whether it was simply that the Asus NT-66U router we’d bought five years ago had gotten old, had some intermittent hardware failure or simply had never been designed to handle having dozens of devices connected to it all day every day, I never determined.

As I mentioned when we set out to replace the router, there was a time when I would have enjoyed the technical challenge of debugging and remediating an old router. The wifi solution at our last house, in fact, was a series of Linksys devices running DD-WRT as a series of daisy-chained repeaters. But between having a young child and just generally finding other things to do with my time as I’ve gotten older, at this point in my life I want to expend essentially zero effort on networking. I want it to Just Work, out of the box. And if it doesn’t, as was the case with the aging Asus device we used, I’ll find something that will.

All of which explains how I ended up asking for recommendations for networking gear on Twitter:

Based on anecdotes and random reviews of networking gear I had seen, I expected the advice to break down into roughly two categories: hard core geeks pushing Ubiquiti, with everyone else recommending some combination of mesh offerings from Eero, Google Wifi and so on. And with a few exceptions, that’s what I got.

Which wasn’t ideal, because as much as Ubiquiti’s reputation made its gear attractive, just the thought of stringing CAT-6 lines everywhere and having to fine tune complicated network settings was exhausting. A simple, cable-free mesh setup was much more in line with my near total lack of willingness to invest time in our home networking solution.

The obvious solution to this dilemma was to go with Ubiquiti’s mesh system, Amplifi: I’d get the simplicity of mesh with the reputation of Ubiquiti. Unfortunately, however, most of the reviews of their hardware like the one from Ars here noted that Amplifi’s mesh points are extremely attractive for toddlers to take apart and destroy. Which is relevant, because we have a toddler.

Initially, then, it looked like I was going down the Eero or Google Wifi routes (Orbi got good write-in recommendations as well), though with reservations about the future of both hardware lines if for different reasons. But then, in a bit of well timed social media outreach, Ubiquiti informed me that I didn’t actually need the mesh points.

When they subsequently DM’d me to ask whether I’d like some gear to test, then, I said hell yes. They sent over two MeshRouters, and what follows is a discussion of how that’s gone.

IMG_20180117_101801 (1)

The Net

The tl;dr version is that the Amplifi gear is excellent, and I recommend it without reservation. Basically, with one exception which I’ll discuss, I never think about our network. Given the degree of traffic it handles – which is a lot for a home, as virtually all of our media (movies/music/tv) is streamed and we have a non-trivial number of smart devices from the aforementioned Nest to WeMo on network – its performance has been exemplary.

In short, while kid duty has had me up in the middle of the night many, many times since we switched over to Amplifi, the number of times I’ve had to attend to a broken network connection has been zero.

Which is all I could ask for.

Our House

As houses go, ours is not that big clocking in a bit under 2,000 square feet. The old Asus router could just barely reach across it, but the connection quality at the far end of the house wasn’t great. So while a three device system likely would have been overkill for us, placing one MeshRouter at each end of the house blankets the entire house in a solid connection. This means, for example, that we can run at outdoor Nest cam at the end of the house furthest away from the internet drop – this would have been impossible with the Asus.

The reach of these devices also means that basically our entire yard – front, back and sides – has internet access. Which is pretty great when I’m loading the woodshed or cutting the lawn; I can stream whatever I want with impunity and without worrying about a stuttering internet connection or reverting to my more expensive and slower mobile data plan. It’s also nice to be able to download the latest podcasts from the driveway at broadband speeds before I head into the office.


Installing the Amplifi system was very straightforward with three exceptions, one which was Ubiquiti’s fault and two of which were mine.

The problems I caused were:

  1. That in conjunction with our replacement of our network setup, the request had been made by a family member who shall go unnamed to change our network ID at the same time. Which was fine as far as the Amplifi setup went, but that change subsequently broke literally every IoT device we had. I’d expected Nest, at least, to handle network changes more gracefully, but I highly recommend against changing your network setup if you have any alternative, unless you enjoy pulling smoke detectors off ceilings and cameras off walls. This has nothing to do with Amplifi, obviously, but in case you get a new networking setup and are contemplating making a change to your network: don’t.
  2. For reasons I can no longer remember, I set our NAS device up with a static IP at some point in the distant past. The problem was that the out of the box IP range for the Amplifi system didn’t match our old router, so the NAS was requesting an IP address that was unavailable. The good news was that as soon as I figured out what the problem was (which took me a lot longer than it should have), that Amplifi made it easy to a) change the default range and b) assign that device the IP address it required. Still, do yourself a favor and don’t set up static IPs if you don’t need to.


The only issue with Amplifi’s setup was with the second MeshRouter. Setting up the first device was idiot proof, and I had it online within a minute or two. Convincing the second MeshRouter that I actually wanted it to be a MeshPoint, however, was a bit of a clunky process. The UI could use some work in that department, but if you’re using the standard MeshRouter plus MeshPoint system, it won’t be an issue for you.

All in all, discounting the issues in which I shot myself in the foot repeatedly, Amplifi delivered the low friction install experience I’d hoped for.


With the Asus, I used to have to remember to login in to the web console and check for hardware updates. With the Amplifi unit, every so often the display on the router has a notification that an update is available, along with a button to install it. One click, the firmware upgrade is applied and it goes back to being unnoticeable.


I’m of the opinion that most routing equipment, and some of the newer mesh systems in particular, are aesthetically mediocre at best and legitimately bizarre at worst. But while I’d definitely lump the Amplifi MeshPoints in that category, the MeshRouters are pretty easy on the eyes. Simple, clean cube design with a highly quality LCD display that doubles as a clock – and that you can configure to power down at night. It’s not exactly Bang & Olufsen in terms of its appearance, but the Amplifi gear is a clear step up from the old Asus, Linksys, etc. routers you might be used to that were oddly shaped and sprouted antennas like a hedgehog.



The Amplifi gear is installed via and managed with their mobile app, available on both Android and iOS. It’s generally well designed, and distills its telemetery down to the actual questions you’d typically want answers to about your network: is the internet up or down? What is its speed? How many devices are connected? Is the mesh network up? And so on. It even includes a built ISP test you can use.


If you’re motivated, the Amplifi system gives you some useful management abilities: you can pause the internet across all devices, or on a device by device basis for example. But apart from the aforementioned NAS IP debacle and cutting our DNS over to Cloudflare’s public network, I haven’t had to tinker too much with the network because it just works.


I’ll leave this question to reviewers who can benchmark the gear against other comparable systems, but I can say that we’ve had zero issues with performance in spite of some heavy network usage requirements. Whether it’s streaming high definition video down from YouTube TV or up to Nest, so far, it’s been so good.


Apart from the MeshRouter as MeshPoint setup complaint above, the only other real issue we’ve had has been with hand-offs between access points. For fixed in place devices such as our Nest cams, this isn’t an issue. But if you have a laptop streaming video and walk from one end of the house to the other, the connection is likely to stutter if not drop entirely as you’re handed off from router to meshpoint or vice versa. That scenario isn’t a particularly common occurrence for us, but it’s not unheard of either so a solution would be welcome.

That’s the bad news. The good news is that Ubiquiti is aware of the issue, and while it’s to some degree dependent on hardware providers such as Apple, the early returns from its updated firmware have been promising.

But do understand that you may experience some connectivity issues if you’re constantly roaming between different access points.

Where to Buy

If you want a system in a box, I’ve got an Amazon referral link for you right here. You can also buy the MeshRouters straight from Ubiquiti here.


Ubiquiti, as mentioned above, was kind of to give me the two review units free of cost.


What’s the Deal With YouTube TV?

The last time I had a cable subscription was a decade ago. Which I bring up only because of the context, not as an attempt to claim any moral superiority. We have a local Netflix-like setup in the house, and over the last decade I’ve wasted just as much time as the average cable customer watching the kind of found footage and monster movies that don’t generally end up in theaters.

So no, I don’t think I’m better than you.

I haven’t missed cable for the most part. As with most cord cutters, live sports has been the biggest problem. I’ve found work arounds for most of it, but they’ve all come with significant limitations. I had to use a paid DNS service that fooled geo-lookups, for one, which was compatible with some but not all of the packages I subscribe to. would work but Netflix wouldn’t, that kind of thing. I was generally limited to watching on a laptop as well, and sometimes a 12″ screen just isn’t the right medium.

All of which helps explain why I’ve been checking in every so often on cord-cutting packages like Playstation Vue or Sling TV. The idea of having an internet only TV package that was cheaper than cable was attractive, but every time I looked there was some critical limitation. As was the case when I first found out about YouTube TV, with its crucial NESN (AKA the station that Red Sox games are on) availability in October. After becoming intrigued by YouTube’s cable-ish package, not least because standard YouTube had become a staple of my viewing habits, it turned out that the service wasn’t available quite yet up here in Maine.

On December 11th, I received notification that the service had launched in Portland.

Cut to a few minutes later.

We’ve been subscribers ever since. When it’s come up in conversation, the number one question I get is “how is it?”
The answer is: it’s great. I like it, but since my judgment with respect to TV is questionable, it’s probably more important to note that Kate likes it as well. She’d prefer it have HGTV I’m sure, but the ability to watch the Olympics for the first time in years has been a big win. As for me, I’ll be able to watch the Red Sox on TV in my house this weekend. Everything else – even the incredible Blue Planet 2 from BBC America, or the amazing recent news that MLB Network will be added to the roster soon – is gravy.

If you’re curious about the details, read on.


At $35 a month, it’s a pretty easy sell. NESN by itself was a $12 extra the last time I subscribed to cable, and as a vehicle to be able to watch sports and series not available on Netflix yet (e.g. Blue Planet 2) it’s a fair value, if not the lowest cost option. Even at $40, which is what the price will go up to in March, we probably wouldn’t think twice about it. It’s likely to pay for itself just in the time I don’t have to spend tinkering with DNS settings to get the sports I want.


At 50+ channels now, YouTube has probably 40 more than I’ll ever need. But more realistically, with the recent addition of Turner Channels such as TNT, Adult Swim, TBS, CNN, and Cartoon Network their last obvious basic cable hole was plugged. Most people will find their basic needs met, whether that’s local channels, live sports or movies – no HBO yet, though Showtime is available as an add-on.

Device Support

Something the service was dinged on early, this has not been an issue for us. We used the native Samsung integration until the Roku client was available, but with support for Android, Chromecast, iOS, LG/Samsung TVs, Xbox and now Apple TV and Roku, most of the common options are covered. I haven’t watched a lot on my Pixel or iPad too much yet, but I can guarantee that those get a workout come baseball season.

Family Sharing

YouTube TV supports 6 accounts but only 3 concurrent streams, so that may or may not work for you given the size of your household. For us, it’s more than enough.


Picture Quality

I’ve seen a lot of debate (for example) around the framerate and resolution of YouTube TV streams, but what I can tell you is that we have a brand new 50″ 4K Samsung LCD, and the picture is excellent. Streams are consistent and don’t buffer for us, either.

IMG_20180120_212004 (1)


The DVR feature is simple to use, and works perfectly. Search for a show, or team in my case, and press the “+” button. That’s it. Whenever and whatever channel the Red Sox play on, this will record it.
Even better? The cloud-based DVR is unlimited, so you never have to worry about maxing out shows. This makes it easy for me to record things like “The Twilight Zone,” which aren’t a priority but that I’ll watch from time to time.


While the user interface is a bit different than standard cable, it’s a lot easier for me to navigate. There are just three tabs:

  • Home: This is a mix of recommendations, shows to resume watching and high level categories like News or Sports.
  • Library: This is the DVR, with everything you’ve recorded.
  • Live: This is the equivalent of your channel browser on cable, and lists what’s on currently and up
The only real usability issues I’ve had came when we were using the Samsung’s native YouTube TV interface; the base-model TV we got came with a base-model remote, and it wasn’t always intuitive how to pause, rewind or fast forward. But that’s Samsung’s problem, not YouTube’s: once we cut over to the Roku, everything’s been seamless.

Less obvious but no less important than the functional usability, in my opinion, are the aesthetics. From the font to the layout, YouTube’s TV package is attractive and a genuine pleasure to use; whatever they’re paying the designers, it’s not enough.


As much as I like it now, there are ways that YouTube TV could be even better. Note that number one on this list would have been adding MLB Network, but that’s already in the works.

  • PIP on the iPad:
    The YouTube TV app for the iPad is, in general, excellent. Its one primary shortcoming is the lack of support for Picture-in-Picture (PIP). One of the things that makes the latest generations of iPad usable across a wider variety of scenarios is its improved support for multi-tasking. Netflix, for example,
    allows you to play media while using the iPad for something else. Currently,
    the YouTube TV app doesn’t offer this. Would love to see the ability added.
  • Channel Filtering:
    Back when I had DirecTV, you could filter their hundreds of channels down based on the ones you actually watched. Which is how people who came over to my house were presented with a channel lineup that consisted of less than a dozen channels including ESPN, NESN, the Cartoon Network, Discovery and the History channel (this was before the latter turned into the Ancient Aliens channel, mind you). I’d like to be able to do that on YouTube TV. It’s not likely that I’m ever going to watch Fox News, so I’d rather not waste UI space on it.
  • Reviews: Given that YouTube TV is offering me recommendations, it’d be nice to be able to give them explicit rather than implicit signals about what I like versus what I don’t. Even a thumbs up/down button, as on Netflix, would work.
  • Offline Caching: As far as I can tell right now, while the cloud DVR is infinitely capable, there’s no provision for offline storage. Unlike Netflix, where I can store a video ahead of time to watch on a plane, YouTube TV isn’t an option for me offline.
  • Better Integration with Standard YouTube: Given the branding and origins of the service, not to mention the growth, it’s interesting to me that standard YouTube isn’t integrated to the TV service outside of “YouTube Red Originals,” which don’t interest me much. I’ve got a growing number of standard YouTube channels I watch regularly, mostly woodworking and DIY-related, that I’d definitely watch on the service if I could. My guess is that one of the potential concerns is what lower resolution user-filmed content would look like on large screens, but I’d like at least the option to even if the picture quality suffers.
  • More Sophisticated DVR Filters (AKA the Simpsons Filter): This is nit-picking to be sure, but while the existing DVR filters are excellent, as mentioned, I’d love for them to be a bit more granular. In my case, for example, I’m a big Simpsons fan, but I’ll only watch through Season 14. The show lost me after that, so I’d love to be able to tell the DVR to record the show so I had them, but to skip the latter seasons where the quality fell off.

The Net

If you’re looking for an easy-to-use alternative to cable and are willing to pay a slight premium for features like time-shifting via DVR, I’d definitely recommend giving YouTube TV a look.

The Problem With Heat Pumps (Or Why We Won’t Buy One Yet)

2014-05-21 09.50.13

A year ago this month, Kate and I bought our first house together. Ultimately we’re planning for a very different configuration, but at the moment it’s a pretty standard single floor ranch. Though the original part of the house dates back to the seventies, we were pleasantly surprised when our inspection revealed that the property was both well insulated and included a furnace of recent vintage – albeit one that ran on expensive fuel oil.

The big problem with these two findings is that neither was true.

This past winter was hard on everyone, and compared to many, we don’t have much to complain about. So let me complain about the size of our heating bills. First, because the furnace was (theoretically) current and therefore efficient. Second, because we had locked in at a price that was $0.20 less than what heating oil eventually peaked at. Third, because we had invested ahead of the heating season in Nest thermostats to help maximize our savings. And last, and most importantly, because we keep our house cold. Not as cold as these guys, but cold. During the week, our program for the Nest is simple: we keep the house at 50 degrees except for an hour in the morning and three or four at night – when we splurge and bump the heat up to 60. Weekends are a little more liberal for obvious reasons, but we were in the top 25% of Nest users every month of the winter in terms of relative efficiency, and the top 15% for two.

While oil is expensive then, we didn’t expect to be paying hundreds and hundreds a month to keep the house just this side of a meat locker. In spite of the horrific nature of the winter.

Which helps explain why we engaged the services of an energy auditor (DeWitt Kimball out of Brunswick, whom I highly recommend). He delivered the first piece of bad news: our adequate insulation situation was considerably less adequate that we had been told. Attics, for example, have a recommended R factor of 40+. Ours was, charitably, in the high teens. Which doesn’t factor in the other various insulation failures we have in a few older windows, the basement door and so on. According the blower test, our house places in the top third of “leaky” homes. We were paying all season, therefore, to heat the outdoors. Which we’ll be addressing by insulating. Heavily.

The far more expensive problem, however, was one we discovered prior to our audit. The oil furnace we thought had been installed in 2005 actually dated – by serial number – to 1992. Our primary heat source, in other words, was 22 years older than we had believed. 22 years less efficient.

The auditor’s recommended solution for heating issue was to keep the oil furnace in place, but to complement it with a “ductless mini-split heat pump,” which are generically referred to as heat pumps or heat exchangers. Popular in both Asia and Europe, these systems are hyper-efficient because they don’t actually use energy to create heat, they simply move it from one place to another – much as your refrigerator does. If a propane furnace is 80 something percent efficient and an oil furnace 90ish (due to the nature of the two fuels), a heat pump is closer to 250% efficient. Couple that with the fact that electricity in Maine is much cheaper on a relative basis than fuel oil – and potentially can be generated on premise were we to invest in solar – and the heat pumps appeared to be a perfect solution. Oh, and you can reverse the direction of the heat exchange and use them as air conditioners in the summer.

The most obvious downside of heat pumps, the fact that they perform less efficiently as the temperatures decline – eventually ceasing to function well below zero – would be a non-issue for us as we already have an auxiliary heat system in place for the few days we see real, deep cold. For most of the winter when the temperatures are 20 degrees fahrenheit and above, we could heat the house using the cheaper and more efficient heat pump. For the day here or there that it got seriously below zero, we could fall back on the oil furnace.

The more I read about heat pumps, the better they sounded. One guy on Martha’s Vineyard heated his house for a year for the grand total of $250 – a fraction of what we paid per month. Another from Presque Isle, almost as far north as you can go in Maine without being in Canada, estimated that he’d save $1,000 a year.

And yet there’s effectively no chance we’ll invest in the technology this year. Why? Because their interfaces are entirely proprietary.

When doing the initial research, I assumed that if there was a problem integrating heat pumps into our existing infrastructure, it would be with our two Nest units. But as it turns out, Nest is more than capable of working with heat pumps as well as an auxiliary heat source like our oil furnace. Called “Heat Pump Balance,” it essentially allows you to use the heat pump until the outside temperature renders it inefficient, then kick in a backup. Perfect.

Except for the fact that, as nearly as I can determine, the best and most efficient heat pumps – manufactured by Fujitsu and Mitsubishi – cannot be integrated into existing thermostat systems, Nest or otherwise. This was the response I got from Fujitsu when I asked about whether their system could be integrated with the Nest or even a standard thermostat:

Our systems can only work with our controllers, they cannot be controlled with the Nest thermostat. At this time there is no way to connect it to the Nest unfortunately.

Essentially, these heat pumps have to be installed as a completely separate system, one entirely independent of your existing HVAC infrastructure – Nest or otherwise. The only way to control the devices is with specialized equipment supplied by the vendor. Which means that we would have two independent, unintegrated heating control systems. Honeywell customer support described the situation to one Mitsubishi customer as follows:

The MIFH1, although built by Honeywell, uses a proprietary Mitsubishi communication protocol to translate and transfer commands between the Mitsubishi equipment and the RedLink enabled products that can also be used with it.

In other words, Mitsubishi’s integration with Honeywell’s Redlink system is a one-off, non-standard connection. Unlike virtually every other piece of HVAC equipment you could buy – air conditioners, gas/oil/propane furnaces, etc – heat pumps cannot leverage standardized thermostat connections. Even if said thermostat, like the Nest, has built-in, native support for heat pumps.

It is somewhat ironic that the most state-of-the-art, technologically sophisticated heating system currently available is unable to integrate with something as basic as a thermostat, but that is the current reality. Which means in turn that as a potential customer, I’m being asked to invest thousands of dollars in a product that cannot be controlled remotely, cannot be leveraged in conjunction with other heating systems, and can’t report telemetry back to somewhere I could use it.

But maybe the real irony is that I won’t invest in these technological marvels for lack of such basic functionality. Even if I could make two independent heating systems work together, I’d be kicking myself next year, or the year after, or whenever Fujitsu, Mitsubishi et al figure out that people want their heating and cooling systems to a) be aware of one another and b) work with each other. It may cost us in the short term with higher heating bills, but it certainly beats paying for heat pumps now, and then again later when the vendors have seen the light and let them work with our Nest units.

In the (hopefully) likely event that Fujitsu and Mitsubishi eventually see the light, my message to them would be simple.


Becoming a Sonos Household

2013-12-27 17.56.38

For years now I’ve been hearing from family (my brother) and friends about the merits of the Sonos system. The advocacy isn’t quite Tempurpedic-level cultish in flavor, but it’s close. Give it few years. None of the wide-eyed recommendations made much of a dent, however. Fifteen years ago I’m confident that I would have had a strong opinion on the Sonos, positive or negative. In those days, I was a wanna-be audiophile. It was bad enough that I was a wanna-be in the financial sense – I didn’t have the money for the real high end gear, which you can generally tell apart by how ugly and primitive it looks. Worse, I was a pretender in ability.

Because I loved music then as I do now, I was heavily invested in the idea that I needed it rendered back to me perfectly. I researched and bought the best equipment I could afford at the time in an attempt to ensure that studio records sounded as close to when they were recorded in the studio as possible. Live shows, which I have a lot of, as if I was there. Marantz receivers, Paradigm speakers, a Hsu Research subwoofer – it was a lot of money at the time.

There was just one problem: I couldn’t really tell the difference between good gear and bad gear. Which meant that I was paying a premium for quality I couldn’t objectively appreciate. This slow, reluctant realization was triggered in part by my transition to digital media: after ripping my CD collection to (ludicrously small) hard drives in the late 1990’s, it was only natural that I began listening to music from the computer rather than my home theater system. And while I’d upgraded the computer speakers, they were not remotely comparable on paper with my home theater setup. But just like the camera you have with you, they had the distinct advantage of being at hand.

If you’re interested, there are lengthy debates to be found on the internet about a) whether the physiological design of the human ear is capable of the differentiation audiophiles believe that it is and b) whether they can reliably tell the difference between equipment costing thousands of dollars from equipment that costs hundreds, if that. Personally, I’m not interested. The only thing I need to know about audiophiles is that I’m not cut out to be one.

Which helps explain, to get back to the original subject, why I wasn’t terribly interested in the Sonos in spite of all of the enthusiasm. I stopped caring about audio equipment over a decade ago and didn’t much intend to start back now. Then my brother got me a Sonos for Christmas.

Specifically he got me a Play:3 and a Bridge. While I’m sure Sonos will tout the quality of its speaker systems, the real calling card is the Just Works quality of the network integration. You connect the speaker to your network, or in my case the wireless bridge, and you can play your music, along with a variety of third party services, wirelessly with zero effort. Even better, it’s modular. You can set up Sonos devices in various areas – living room, kitchen, TV room – and push any combination of local media and services such as Pandora to any combination of those rooms. The appeal is obvious, even for someone who’d given up on stereo tech. Convenience, as ever, is a killer feature.

Even now I’m not sure I would have invested in Sonos without the push from my brother; the upfront capital costs of the entire system would be discouraging, as the bulk of our available capital is currently allocated to home improvement. But with the initial purchase out of the way, I can easily see us expanding from the dining room (where our unit sits now) to the kitchen, living room, bedroom, maybe even the basement.

Some of you are probably reading this and thinking, “Why pay Sonos when I can just connect a home media server to my home theater system via an RC cable?” Good question, as that’s exactly what I used to do. For me, however, the answer is time. Given time and effort, I could duplicate the functionality, or most of it, using off-the-shelf stereo components, an old server, VNC and some ethernet cable. But then again, I pay Dropbox so that I don’t have to (poorly) replicate it using cron jobs and rsync scripts. Sonos isn’t the only way to push music around your house, but it’s almost certainly one of the easiest.

Once upon a time I had more time than projects to tinker on. Today, that ratio has flipped so completely that my vacations are spent working on the projects that I can’t get to when I’m working. As such, the Sonos has a lot of value to me. Might be it does for you.

The Good

  • Clients: Sonos doesn’t support, at least as far as I can tell, a Linux client, but with Android, iPad, iPhone, OS X and Windows, they have most of the bases covered.
  • Setup: Setup was surprisingly easy. Plug the Bridge into the router, press a button and it’s configured. The Play:3 was even more interesting. It somehow got itself onto my encrypted wireless network without my ever providing it with the password. I assume it was assisted either by the Android device I configured it with or the Bridge but either way, I’m curious.
  • Sound: As mentioned earlier, I’m no audiophile, but the sound quality is very good, even at higher volumes.

The Bad

  • Library Updating: The process of integrating a music library can be less than intuitive, particularly if – as I do – you store your music in a separate directory from your iTunes library file.
  • NAS integration: Sonos does support integration with NAS devices, but it’s a heavily manual process. Would be nice to see them work with vendors like Synology who supplies the box we use here.
  • Service Portfolio: While Sonos has covered a lot of the bases here with Audible,, Pandora, Rdio, Spotify, etc, there are some obvious holes like Google Play (though this app is a decent workaround) and iTunes Radio. Based on the politics involved there, it might be a while.

The Ugly

  • User Experience: Literally. Sonos does a nice job with the out-of-the-box user experience, but their UIs are, in general, not attractive, and usage is a bit awkward.

Three Technologies That Impacted Me Day to Day in 2012

Every year is a different mix of technologies. Some are here to stay, others are not long for the world. But for future reference, if nothing else, I wanted to snapshot a few things that were very useful for me in 2012, where useful means they impacted my day to day life. I’ve also tried to focus on lower visibility items: it would be true to say, for example, that the Nexus 7 I acquired to replace my Xoom impacted my life day to day, but there are hundreds of articles about tablet usage. Same with cable cutting and the Roku.

This is a list instead of things you may or may not have heard of, but might prove useful if they’re new to you. The latter two are Android-centric, but the first is available to anyone. Enjoy.



Of the technologies I started using in 2012, none had a wider impact on our house than Plex. For a variety of reasons ranging from cost to the fact that I dislike TV in general, Kate and I have never had cable in the house. Most of her shows are available online via Hulu and other channels, and the only thing I care about – the Red Sox – is available via the radio. And, when I’m traveling.

But watching movies together was always a trial. Either we had to first decide on and then hunt down a DVD, which were stored haphazardly across four or five different nylon sleeves, or it was Amazon Video or Netflix – both of which posed problems. First of all, we live on an island where we have only one broadband provider – Fairpoint. Who isn’t very good. Instead of being “27 times faster than dial-up,” our connection at home was regularly more like two. The other issue is that our TV’s old enough to not have Netflix built in, so we had an old Mac Mini plugged in for the streaming services. Which meant using a keyboard and mouse to queue up movies.

Ridiculous, I know. Anyway, to solve the latter problem we plugged in a Roku, which worked perfectly. But we still had the bandwidth issues. Also, the selection on Amazon and Netflix was imperfect.

Enter Plex. A really marvelous and underappreciated piece of software, Plex is what I’d wanted Google TV to be: a free as in beer home media server software package that does a number of things. First, it provides a clean, 10 foot navigable interface to media – movies, music or TV – that you load into it. Second, it will automatically retrieve metadata such as movie posters, actors/actresses, genre and so on in the background. Third, it will let you organize it into collections: we have one for movies that we find mutually acceptable, for example (it’s a very short list). And lastly, it will stream this content to computers, phones, tablets, TVs, Roku boxes or anything else connected to your network, transcoding it as necessary on the fly. It will even let you stream it remotely. When I’m in Brussels next week, Plex will let me watch movies off of my home server. Isn’t the future cool?

The developers in the audience might also be interested to know that Plex’s new interface is Bootstrap-based, and very well done.

If you’re interested in Plex, here are two posts on how I built what we call sogflix – our local Netflix equivalent: “How to Build Your Own Personal Netflix” and “How I Rebuilt Our Entertainment System Using Plex and Roku.”

Google Now

Over the long term, Google Now is likely to have the most impact on my day to day life, because it’s about putting my data to work for me. And while Google’s really only scratched the surface with Now, even today, the application is enormously useful.

When I’m flying home from Baltimore, for example, it knows from my Inbox what time my flight is, and its status. It also knows from my search history that I like Of Love & Regret – where Ryan and Leigh of Monktoberfest fame work – and thoughtfully provides me with directions there as well as traffic time.

2012-12-30 10.13.23

If I’m scheduled to meet a friend for a beer, it provides me directions there and will even tell me when I have to leave to be on time.

2012-07-26 17.08.58

Google Now also knows how long it’ll take me to get to work and what time the Red Sox game is on.

2012-08-20 13.06.32

From my travel itinerary, Google Now pulls up the weather forecast at each stop along the way.

2013-01-07 05.23.45

If I’ve ordered a package, it will track it for me and keep me informed of how much I’ve walked or cycled over the past month (for the record, I only had my phone for a week that month).

2012-11-01 15.55.13

The only thing I need to know to use Google Now is how to turn it on. There is nothing else to learn. Google Now is easily the best new feature to Android in the last release, and possibly since it launched. While the privacy implications will undoubtedly scare off some potential users, at this point I’m as comfortable with Google on that score as I am anyone. And the ability to outsource weather, package tracking, flight status, day to day schedules and so on to software that can leverage my data alongside that from external sources is quite compelling. Day to day, Google Now is a godsend, and projecting forward a few years it’s likely to become only more useful.



At some point over the last year, my friends gave up texting me and started texting my wife. The latency from my replies was just too high.

Not too long ago, text was actually a great way to reach me. Quick, efficient and not an overrun wasteland like email. But once I got a Nexus 7, it replaced my phone for casual browsing and other tasks at home. Which meant that a text directed to me would wait until I picked up my phone. Which could be hours, or in rare cases a day, later.

So, suboptimal if you’re looking for a quick response. Hence the texts to my wife. A few months ago, however, I was directed to the Android app MightyText. Billed as iMessage for Android, it essentially connects SMS to the desktop and a tablet. When I get a text now, the Chrome extension on the desktop and my Nexus 7 will pop up a notification, and I can reply to the text quickly. Using a real keyboard, no less. MightyText will also notify you when someone’s calling your phone, and display the phone’s current battery level.

But the ubiquitous and device independent access to SMS is the big feature for me. Now I get texts when they come in, rather than hours later. Which may in time lead to less texts from my friends to my wife – we’ll see.

In the meantime, MightyText is a free app. The handset application is on Play here, while the tablet version is here.

Disclosure: There's nothing to disclose. None of the above products were built by clients, nor do I know people personally working on these projects.

My SquareTrade Experience

Because I wasn’t willing to spend north of $600 on a handset, it took me a few months to work out an upgrade scheme by which I could move from my aging Nexus One to its newer cousin, the Galaxy Nexus. In March of this year, I bought an unlocked Galaxy Nexus from Amazon, then used my AT&T subsidized upgrade to secure a Galaxy Note. For reasons I still don’t understand, the Galaxy Note sold for more than the Nexus cost, so in effect I bought the Nexus for a subsidized $299 cost in spite of the fact that AT&T never offered the device.

Anticipating, however, that replacing the phone in the case of breakage would likely be costly and complicated, I purchased an after-market warranty from SquareTrade. For $69.99, I bought 12 months of coverage against accidental damage. Which I engaged two weeks ago when my Nexus became entangled in a cord, dropped on a granite counter and cracked the screen in three places. Here’s the good, the bad and the ugly of my SquareTrade experience.

The Good

The good news is that SquareTrade was able to repair my Nexus perfectly, with no questions asked. The phone was returned to me in exactly the condition I sent it to them, except for the brand new display. We’ll see how the repair holds up over time, but from the initial inspection the work was done correctly.

So ultimately, I got exactly what I wanted: my phone restored to its pre-dropped state. As a bonus, the logistics of the return were straighforward. I filed a claim online and they processed it same day. The claim return included a shipping label; after packing the unit up I dropped it off at the local UPS store and walked away.

The Bad

The bad news isn’t SquareTrade’s fault exactly, but the economics of the warranty are less compelling than I’d hoped at the time of purchase. Since I bought my Galaxy Nexus, prices for the handset have come down significantly. With the Galaxy Nexus’ replacement the Nexus 4 now out and selling for as little as $299, the cost of an unlocked Galaxy Nexus is around $200 today. Compared to that replacement cost, the SquareTrade warranty, which came to $168.99 including the $99.00 deductible, represents a marginal savings for a replacement handset. For a mere $130 over the warranty cost, I could have upgraded from the Galaxy Nexus to the Nexus 4.

The Ugly

The real problem with SquareTrade for me, however, was the delay. Although I didn’t realize it at the time, the shipping label provided to me by SquareTrade was for UPS Ground service. Because the unit shipped from Portland, ME and was headed to Los Angeles, the transit time was a full week. The unit was dropped off at the Portland UPS store Monday the 15th and was signed for in LA Monday the 22nd. It then took SquareTrade three days to process my claim; the unit was shipped backed to me Thursday the 25th. To SquareTrade’s credit, the return shipment was next day air, and it’s not SquareTrade’s fault that while UPS promised to deliver the package Friday it failed to due to “adverse weather conditions” on a day that was 48 degrees and partly cloudy.

Even in a best case scenario, however, my total time without the handset would have been ten business days. I was fortunate to have my Nexus One to use as a replacement, but many users probably don’t have second phones readily at hand. In which case they can be expected to be without a phone for a period of days, best case. For most people, that delay will be unacceptable.

The Net

In the end, SquareTrade lived up to their promise: they repaired my phone for the agreed upon deductible, and my Galaxy Nexus has been restored to its prior condition. It’s not clear, however, that the extended duration of the replacement process was worth the marginal cost savings the warranty afforded, though the economics would admittedly have been more in SquareTrade’s favor if I’d broken the device earlier in the contract.

The net is that those evaluating a potential SquareTrade warranty should consider carefully the replacement cost of the device, both at the time of purchase and over the life of the contract. In my case, with Google making their handsets available directly for half the cost that I paid originally, I’m unlikely to purchase another SquareTrade warranty moving forward for a phone.

If you lack realistic service options, however, and the cost of your device justifies repair, SquareTrade may make more sense for you.

How to Build Your Own Personal Netflix

Back in April, I wrote about our decision to update our antiquated home media setup with a Roku box and Plex software, among other pieces. The motivation was simple. As non-cable subscribers, we were looking to bring our multimedia setup into the 21st century. And with Plex specifically, we wanted to hedge against the varying variability of NetFlix titles as well as our own problematic bandwidth (the downside of living on an island). Thus, sogflix – our local NetFlix equivalent – was born.

After seeing it in action, a few people have asked how it’s done. So here’s what you need to know to duplicate it in your own home. All Amazon links are affiliate links, FYI.


Here’s what I’d use to make it all work. Note that this list was updated in January 2018 to account for discontinued hardware and cheaper, larger capacity drives.

  • Handbrake (free)
  • Plex Media Server software (free)
  • Plex Media Server hardware (can be any Linux, Mac or PC hardware with the exception of the Apple TV)
  • Roku (Premiere, recommended b/c of its ethernet port – $65.69)
  • Storage Unit (RAID configured)
    1. Synology DS218 ($249)
    2. (2) Western Digital (Backblaze recommended) 6 TB 7200 RPM ($219.99)
  • Ethernet cable
  • HDMI cable ($2.50 @ Monoprice)

Convert Your Media

The first step in building a digital library is, of course, digitizing the library. This is the most tedious and time consuming part of the process, depending on the hardware you have available. My workstation will convert a movie in about twenty minutes; my Thinkpad takes closer to two hours. It’s fairly easy to run as a background process, however, converting media while you work on more interesting subjects.

With the caveat that you should only convert movies that you are legally entitled to – consult your local laws – the simplest way to do this is via Handbrake. After a great deal of experimentation and help from a few Plex employees, I’ve come up with settings that Plex is able to serve and Roku is able to display. For the long version, see the site here. The short version for me is starting from the Handbrake defaults and doing the following:

  • Open picture settings and
    1. Turn Anamorphic from “Strict” to “Off”
    2. Set the Decomb filter to “Default.”
  • Make sure the framerate and video quality are set to constant, and set the latter to 17.
  • Use the following naming conventions:
    • Movies: Title (Year).*
    • TV: Show Title – s##e##.*

The output from this will be movies of somewhere between 1 to 3 GB per, depending on length. Just as an MP3 is a lossy copy of the original, the quality is almost certainly a compromise. I’m not able to tell the digital copy from the original DVD, however, so that seems sufficient.

Hardware Setup

  • Audio / Visual:
    Initially, I was content to use the analog audio/visual patch cable from the Roku to the TV. This is a workable solution if your TV is not HDMI capable. Upgrading to an HDMI cable, however, at a cost of $2.50 from Monoprice yielded immediate improvements in both formatting (menus fit the screen perfectly) and video quality (Plex movies were substantially improved).
  • Networking:
    If you have a pure 802.11n networking setup, you may be able to avoid hardwiring. In my case, my Mac Mini was old enough that its networking was 802.11g which proved to be insufficient, causing local movies to buffer as if they were being streamed. The solution in my case was connecting both the Mac Mini and the Roku physically to the access point (which in my case is actually the fourth in a chain of dd-wrt formatted WRT-310N Linksys routers). Once connected by ethernet cable, video loaded in a few seconds and required no buffering.
  • Storage:
    For non-technical users, set up of the Synology unit may be moderately challenging, but it’s manageable. Those reasonably familiar with storage should have no difficulty, with that the caveat that install media is packaged for Mac and Windows only. Install the hard drives into the chassis as intructed – you’ll need a screwdriver – and then follow the instructions on the software to install the base image on to the Synology, and format the drives. Once set up is complete, create separate directories for Movies and TV. I do not recommend music if you have a large collection (> 10K tracks). Plex struggled to add metadata for my music and performance suffered.

Media Transfer

With movies converted and the storage media prepared, copy the remaining movies to the media using the fastest mechanism possible. The DS212 supports USB 3.0, notably. Movies can all be located in the same directory. For TV, you’ll need to use a [Show Title] [Season #] directory structure, where [] designates a directory.

Plex Configuration

Install the Plex Media Server, available for download here for Linux, Mac and Windows. Once installed, run Plex and select the Media Manger. Within the Media Manager UI, create Movie and TV sections within Plex, pointing the sections at the directories you just created and populated with media. Before you click update to add metadata, you need to make one adjustment. To avoid Plex retrieving foreign movie posters for you media, click the “Metadata Agent Settings” at the top of the Plex Media Manager, then click “Freebase” under Movies. Drag MoviePosterDB to the top, then TheMovieDB behind it. Finally, click “Edit Section” at the bottom of the Media Manager and select “Freebase” as your primary metadata agent.

Then click update and Plex will retrieve a wealth of metadata – movie posters, actors, directors, genre, descriptions and more – for your video. Assuming it’s in one of the databases, of course.

Roku Configuration

Within Roku, add the Plex Channel. If your Plex Media server is running and on the same network as the Roku, it should autodiscover your server and you’re done: Plex Media will now be playable through your Roku. Enjoy! You now have your own personal Netflix.

How I Rebuilt Our Entertainment System Using Plex and Roku


Update: For specific implementation details including recommended hardware and costs, see the follow up post here.

Neither my wife nor myself had cable when we met, and even as a couple we never bothered to subscribe. It’s not really a question of content, exactly: there are several shows she watches regularly, and I’d love to have Red Sox games in the house. It’s more the cost: we don’t really see the point of paying almost a hundred dollars a month for hundreds of channels we’ll never watch. We’re not alone, obviously: more and more people are cutting the cord every year. The downstream impact of this attrition will be interesting – and in the case of efforts to kill net neutrality, horrifying – to watch, because the carriers for us are what they refuse to be: a dumb pipe.

Up until this week, however, our multimedia experience was terrible, especially for a household with one member whose employment description includes “technology analyst.” Aside from watching Amazon or Netflix videos on our laptops, we had an old Mac Mini connected to my old Sharp Acquos 37″ TV. If we wanted to watch something together, then, we manually queued it up on the Mini via a browser, manually expanded the Flash or Silverlight player to full screen and then prayed we wouldn’t need to pause it for any reason. Because that meant getting up and using the mouse or keyboard to pause the video. Seriously. In 2012, we had no effective remote control.

Ridiculous, I know. But I fixed all of that this week. Here’s how.


The first decision to make was Apple TV or Roku (Boxee got no votes from my audience). Either would at a minimum return us to a world containing remote controls, but I had to determine which to get. In spite of the two Mac laptops, the Mac Mini and my wife’s iPhone, we don’t really consider ourselves an Apple household. But more importantly, the Apple TV wouldn’t connect to either Amazon Instant videos or HuluPlus, both of which we subscribe to. Which made the decision easy.

We bought a Roku XD – which will do 1080P HD, but not play games – at Amazon for about two dollars off the devices suggested $79 retail price. Setup was simple enough that I’m pretty sure that my parents could complete it without assistance. A little bigger than a hockey puck, the Roku was unobtrusive and offered us – finally – the ability to watch content from Amazon, Hulu, Netflix and so on without messing around with a keyboard and mouse. For the baseball fans in the audience, it supports, so you can watch any game except those that involve the home team (unless you get creative: think VPN). We’ve added other channels, like CNN and PBS, but for the most part we were watching our usual mix of Amazon, Hulu and Netflix.

By itself, the Roku was a massive improvement over our previous setup. But the inability to integrate music, movies and TV that we had on hand was unfortunate. Then I discovered that Plex had an official Roku channel.


For those unfamiliar with it, Plex is free to download software that serves as both media server and front end. At one point, I tried to go Plex only, approximating the Roku’s functionality, but using it with the Mac Mini’s little two button remote was tedious. Instead, we now leverage as a back end, just another content provider for our Roku front end.

First, I downloaded the Plex Media Server package, set it up on the Mini, pointing it at the local Dropbox maintained music repository and the Google Drive synced video directories. Then I added the Plex channel to the Roku, and had it discover the Plex Media Server running on the Mac Mini. In addition to serving it up, Plex reaches out and pulls down images, descriptions and category information for everything from film to TV. Five minutes later, I was watching a movie off the Mac Mini via the Roku. The net, then, is that the Roku can pull in all of the major content provider channels – as well as anything I happen to have on hand locally.

As an added bonus, Plex has an Android app ($4.99) that will let my Galaxy Nexus stream anything from my Plex Media Server. Like movies, for example:

In theory, I could access the content remotely as well, though our current network configuration isn’t really set up for that now. Crazy, no?


  • One technical limitation I’ve discovered thus far is that the Roku only speaks a couple of video formats natively, so there were some buffering-like pauses as Plex transcoded the video on the fly, but Handbrake should be able to convert the videos I have on hand (it’s actually what Roku recommends) to something Roku’s more comfortable with. If you have a lot of video, however, this may be less ideal.
  • The biggest problem with cutting the cord today is live sports. Think carefully about how much sports you watch, because that might decide you against going without cable. A few leagues like MLB are streaming their games today, but most are not. And even in cases where they are, like MLB, the blackout restrictions – intended to prevent people from doing what we do – are likely to be a problem. I’m fine with it because I have no problem listening to baseball on the radio, but your mileage may vary.
  • The Mac Mini’s wireless card is b/g, so wireless bandwidth does not appear right now to be a limiting factor. But depending on how much video we’re pushing around to different devices, we may have to upgrade our local network of WR54TG dd-wrt formatted access points and repeaters to 802.11n infrastructure, and get a new wireless card for the mini.
  • If we’re really going to do the local media thing correctly, we’ll probably need to get a few new terabyte or two terabyte hard drives (suggestions welcome) at some point for storage and backup. The media right now resides on a single Maxtor 250GB external drive, which I don’t trust.

The Net

In less than a week, our media setup went from circa 1975 to 2015-ish, I think. The Roku’s a more than capable media front end, and the Mac Mini – which otherwise is a terrible, slow machine – gets a new lease on life with its new role as a home media file server. We can now browse virtually every popular internet content source (with the exception of YouTube, which can be played via Plex) and virtually any local media quickly and efficiently via a remote, with playback very straightforward. Or as straightforward as streaming services get.

All of that, for $78 (Amazon shipped it for free), or what one month of cable with a premium channel might cost you. Sure, HuluPlus+, Netflix and so on aren’t free, but even combined they’re a hell of a lot cheaper than your average cable bill. The Roku/Plex experience is, frankly, what I expected Google TV to be, but with Logitech losing a $100 million on that business, its future is uncertain. In the meantime, however, the combination of Roku and Plex is both compelling and cheap. If you’re an Apple fan and don’t use Amazon or Hulu, the Apple TV is a good option, but for everyone else the Roku’s worth a shot.