> ----- > - Give some background of your experience in free and open source > software. Sure. I call it "Libre" now, though, but that's not for the usual reasons which people label as "ideaoligogogical". Thanks to working with Chris from Think Penguin I've become aware that actually, respecting Software Freedom has absolutely nothing to do with "ideology", it's about selling products that are stress-free for the users and just work, pure and simple! If you've ever tried upgrading a laptop over WIFI and been disconnected half way through the upgrade because the new kernel you just installed is incompatible with the older proprietary firmware and you now can't even connect to the Internet to complete the upgrade, you will know exactly why it is better to have a WIFI card that "just works" across a wide (and often near-historical) range of Linux Kernels. So. Rewind: let's go back to... 1992. I was working for PI Technology in Cambridge where we had a SunOS 4.1.3 server and Hummingbird Exceed NFS clients for DOS, and everyone was using Windows for Workgroups. When it came to upgrading to Windows 95, I wanted to use NT 3.51 as my development platform, but the Hummingbird NFS client was a king's ransom. Luckily, one of my beard-endowed colleagues piped up and said, "well there's this thing called Samba..." - so i spent the next few weeks going over every single function in what was then Samba 1.9.15p6, converting them to use the horrible non-ansi-C-compliant sun cc compiler because, for security reasons they would not permit fixincludes to be run on the SunOS server. Now, that experience - of having to go over every single function - led me to explore the code in more depth. Some six months later I submitted a huge patch to Andrew Tridgell, to add the beginnings of decent Network Neighbourhood support. It really went from there, culminating in the NT Domains Logon Client and Server functionality [1], which, hilariously only a few years ago, I actually compiled up using the MinGW compiler as rpcclient.exe and smbclient.exe and ran successfully under Wine, as a way to test a patch to add proper support for NT NamedPipes in Wine (which still has not been resolved after fifteen years - [2]) The key lesson that I learned from the experience of working with Software Libre however has only recently come to light publicly and loudly. I don't know if many people remember the Lead Developer of Gentoo who, many years ago, had forty five THOUSAND dollars of credit card debt and was forced to take a job with Microsoft of all people? More recently there was the developer behind GPG whom people had also forgotten to fund. The more public and serious ones were of course SSL and Bash related major security vulnerabilities, where Corporations finally woke up and realised that they were making all this money but forgetting to actually make sure that some of it went to the teams that were ultimately responsible for the huge profits that were lining their pockets. This is one of the key disadvantages of all "Software Libre" - that there is a fundamental disconnect between the time at which people "get the goods" and the time at which the people *providing* those goods actually receive a financial reward equal to the value of the work that they've done. This used not to be the case: everybody always released the source code, everybody always contributed and collaborated, so it came as a bit of a shock when proprietary software companies started *not* releasing source code modifications to BSD-licensed code. To say that Corporations are acting unethically by not recognising this would be a huge understatement, so I was one of the first prominent Software Libre Developers, who saved businesses world-wide several billion dollars in proprietary software licensing fees, to fully appreciate this disconnect, and I learned it the hard way. Around 2003 I was working on building sites as a common labourer for a while. So, fast forward a bit and we have things like Android: a few years prior to that I had spent a couple of years reverse-engineering HTC Smartphones to get the Linux kernel up and running on it. In one fell swoop Android wiped out the opportunity to bring better Software Libre OSes to a wider range of people (with the associated stress-free issues, privacy issues and so on all left unresolved), by *reimplementing* a huge range of core GPL components, releasing them under BSD-style licenses (which Corporations have repeatedly abused, including Apple), but hypocritically Google did *not* reimplementing the actual Linux Kernel as a BSD-style project because they realised that it is beyond even them. This has resulted in huge confusion in the eyes of companies like HTC, Motorola and many more, who, on reading the BSD-style license in Android, fail to grasp that Android != Linux Kernel. In other words, we know from the experience of Android that not even companies that tell us that they're doing us a favour by working with "Open" source are actually not really interested at all in acting in an ethical or responsible fashion: they just want your money and are prepared to use whatever means is available at their disposal to get it (people forget that Google is an advertising company, not a search engine or an Android-device-selling company). The bottom line is: I began to realise that there was no point sitting here as a Software Libre Developer complaining about how these hardware companies were taking advantage of Software Libre, I was actually going to have to do something about it and set up shop designing and then selling actual Hardware. Then and only then would I be able to receive actual cash at the time of the transfer of physical goods to people, that I would then be in a position to install ethical and privacy-respecting software that "just worked", that respected Software Libre Principles (and the associated hassle-free business implications which are still not very well understood even by prominent leaders in the Software Libre world), and from there I would be able to set up an ethical business fund that actually gave something back to the actual people on whom that business had actually been successful. And, ultimately, do things like fund the creation of a useful and powerful privacy-respecting processor (which won't be based on RISC-V - sorry [3]). > - I understand that you were involved with the Improv Dev Board and > Vivaldi. What have you learned from those experiences that would help > with the EOMA68 line? Excellent question. Ok, first I should outline exactly what EOMA68 is for (not what it is: that can be seen on the standard page [4]). EOMA68 is designed to solve a huge range of issues associated with the cost of ownership of mass-produced computing appliances. Right across the board, all the way from end-users to factories and fabless semiconductor companies it has enormous benefits, which I've outlined in a white paper on ecocomputing [5]. Secondly, I feel it's important to understand what happened so that I can actually *tell* you what lessons I learned. Aaron contacted me after also experiencing the same difficulties with the rampant and endemic Android Tablet GPL violations that were documented by mjg59 around 2012 [6] [7]. Like many engineers looking for a way to leverage the significantly lower cost of mass-produced tablet hardware ($35) by re-programming it instead of being forced to spend $20,000+ in hardware development NREs and thus becoming completely unprofitable, Aaron's team went through a stack a METRE HIGH of tablets that were not only GPL violating but that the factories simply had no idea what source code was, and, worse, because they felt that because they'd spent $USD 10,000 to buy that design (including the GPL-violating binary-only OS image) they going to investigate this thing called "source code" some time just after Hell Froze Over. So against that background, and having been ripped off already by a factory that changed the entire innards of a tablet (because it was cheaper for them) well after Aaron's team had managed to reverse-engineer their OS onto the processor in the older board that the factory didn't understand was really rather important, contacting someone like myself who has a background of sticking to ethical software libre principles seemed like a really good idea. With Aaron's background coming from Project Management in a hardware-related company, the prospect of funding an entirely new tablet did not scare him, and, thanks to the main EOMA68 computer part having been completed and proven, the remaining part (the base unit into which the Computer Card plugs) was anticipated to be much much simpler: high speed signals would be a single set differential pairs (USB2), the whole thing could be done as a simple 2 or 4 layer board, and so on. However as I personally did not have any PCB CAD development experience at the time, we explained to Aaron that it would be best to use a third party OEM in China, and that I would manage them. Now, the company chosen in China was one that I had done some business with before, but I was unprepared for their lack of ability to follow simple, clear and obvious directions. I specifically asked them to ensure that they designed the PCB and the casework simultaneously. I was reassured repeatedly that this was going to be taken care of. Months passed, and when the PCB and case arrived, the PCB would not even fit in the case. They wanted another $USD 10,000 to fix the problem: understandably we said "absolutely not!". They then wanted us to pay them $USD 10,000 before they would send us the PCB CAD files - something that was in no way made clear when we started. So, we had to walk away from the entire investment and start again. Aaron was not very happy with this but was at least understanding. What he was less happy with was that he had agreed to the initial specification (including to provide a maximum of 640x480 camera support) but he had not really fully thought through the implications at the time that he'd *agreed* to those specifications, if that makes any sense. Hardware being what it is, it's not like it can be changed without throwing out the entire PCB and all samples made to date and starting again from scratch, so he wasn't very happy but was philosophical about it, having seen this sort of thing happen many times before. In retrospect, what probably happened is that he discussed the specs with his team *well* after agreeing to them (including the limitations of 640x480 on the camera), got some flak for it, and then tried to blame us. Anyway, what he didn't tell us - but soon became important - was that he did not have a stable, sustainable revenue stream: the entire project (as well as the KDE team) was being funded personally from a previous successful deal. With the tablet not being successfully done first time, he decided (and we agreed) that it would be a good idea to get the proven EOMA68-A20 Computer Card out there, so that people could start developing software for it, and, because I had just worked with a team to create a Micro-Engineering Board (a minimalist board with only 10 components on it - mostly connectors), Aaron would put a team together to create a slightly larger but just as equally simple board design together. Now, this is where the problems really began, because what I didn't understand was that I'd just handed over design control, at a critical strategic time, to a completely unknown team whose ability to work together and to listen to important design decisions was completely unknown. Over several months, the catalog of mistakes that they made on such a simple board with only 12 components on it was staggering. They tried sourcing components from the USA (putting the BOM up to $25 when the target was under $12). They didn't read the EOMA68 specification and designed (and made!) samples that did not comform to the specification, costing them an extra couple of months and a lot more money. They messed up key strategic relationships with suppliers that I introduced them to. Attempts to help steer them through the lessons that I had *already* been through were brushed aside. Bear in mind, this is against a background of Aaron's financial resources dwindling: delays could therefore have a critical impact. The next mistake that I made was to let the level of crowd-funding be set at 2,500 instead of 250. Bear in mind that we had done a successful run of 25 EOMA68-A10 Computer Cards, and sent them out to early adopters. The next logical step is 100-250, so that any problems in the batch production can be detected and sorted before going to the next scale up. In China, these steps are entirely skipped and they test out MOQs of 50,000 units on unsuspecting vendors and customers, presumably on the basis that if it all goes wrong, if nothing else they can always disappear and set up shop elsewhere in the huge country called "China" with its 1.4 billion potential suckers. So, bear in mind, we have had the proven EOMA68-A10 (now A20) board available for some time. We've lined up a factory to produce it (and staked our reputation on the order) - this factory being a China State-Sponsored factory that assembles 20 million mobile phones a week, makes 30% of the world's Desktop PC Power Supplies, and makes the glass for LG's TVs and LCD panels. As in, *all* of LG's TVs and LCD panels. To say that this is an important deal, in other words, for Software Libre to demonstrate credibility and to gain access to an extremely powerful but R&D-limited channel, would be a massive understatement of the first order. The pricing is set at $75 based on the expectation of reaching 2,500 units: there is no wiggle-room here. $12 is allocated for the base unit, $18 or so for the CPU Card (thanks to our large factory), then there's packaging, shipping and so on - it's pretty tight but doable. Anyway the crowd-funding campaign goes ahead (despite unvoiced reservations which I should have made clear and put my foot down on), and we leave them to it. Every couple of weeks we ask "how is it going, what's the numbers like". We do not receive a response. We increase the urgency of the enquiries, and do not receive a response. Now, half way through, the forum has some questions from interested backers. The team that took well over six months to create a PCB with only twelve components on it begin to respond to questions about the EOMA68 standard, answering as if they are authoritative and authorised to make such statements. Just one of their statements made an assertion about future versions of the standard which, if it was actioned, would actually destroy the entire standard and bring it into disrepute by making it impossible for users to rely on it. I had to step in and in no uncertain terms - bear in mind that this is entirely public, on a forum and in the middle of a crowd-funding campaign - not only correct these unauthorised individuals but also publicly reprimand them, in order to protect the standard from being brought into disrepute by their attempts to take control. Towards the end of the campaign we had a meeting and guessed that Aaron was not responding to communications because he was nowhere near the expected target of 2,500. We made a worst-case educated guess that he must have reached 500, and urgently contacted another team who owed us a favour, and asked them if they could take the time to quote us. They said that 500 was a bit small, but that they would contact their factory because their business was successful due to our work with Allwinner in getting GPL compliant source code out of Allwinner. So, again, not only is our reputation on the line with another team, but that team's reputation is on the line not just with us but with this other team as well. The quotes come in just as Aaron finally contacts us, pissed off and blaming us that he's only got 250 orders. Note that that was the amount that I had previously figured would be a good idea to target, but that I had not voiced that strongly (if at all). We tell him that there's a quote of 500 we've received: could he pay for the remaining 250 units and keep them as stock? It's at this point that he finally reveals to us that he's running out of money, so no, he can not pay for the remaining 250 units. Skip forward to last week, and I discover only now - nearly four years later - that after this all occurred, Aaron had been going around telling all of his friends and anyone else who would listen that we had "gone silent" and had "failed to deliver our end of the bargain". There is a common psychological phenomenon where when people wish to blame others, the first excuse that they come up with is the one that they themselves are actually responsible for. So against this background, the question "What did you learn" has a heavily-emphasised answer, "A Hell Of A Lot"! Here's a short list: * Don't trust complex product design to anyone else (I therefore taught myself both PCB CAD design and 3D CAD design). * Don't hand over control of a project to anyone else, in any way, particularly if it's at a strategically critical juncture. * Don't let other people take control of the EOMA68 Standard. Ever. * Don't work with VC-funded ventures. Find someone with a stable revenue stream instead, as a sponsor. * Don't work with unethical people if you have an ethical business model. Ever. It's not going to work. Whilst these may be viewed as "negatives" - lots of "Don't" involved - you have to bear in mind that part of the strategy for EOMA68 is that it should be as inclusive as possible, and that, in approaching an entirely new and completely unheard-of industry initiative, I really did (and still do) want to give people the opportunity to help out. Within that context, then, it was actually very important for me to learn what boundaries should *not* be crossed in permitting other people to help out, hence the reason why these things are listed in my mind as "Don'ts", not "Dos". In retrospect, though, it was a good thing that the Improv board did not go out. The reason is that the EOMA68 standard was not really ready at the time to become a relevant standard for over a decade. It still had SATA and Ethernet on it, which were replaced in later versions by SD/MMC, SPI and USB 3.1. USB 3.1 maxes out at something like 10 gigabits per second and is a general-purpose bus, whereas SATA and Ethernet are fixed-purpose and are slower speeds as well. But, the one piece of lasting damage is the lost opportunity to work with an extremely powerful factory. Our reputation was destroyed by not delivering on what we had promised - a single simple order of only 2,500 units when they usually deal with 250,000 minimum quantities because their assembly lines are so ridiculously enormous. But, on the flip side, we can take the view that given that EOMA68 was not ready, it was better that the deal did not go ahead at the time, because that version of the standard would not have survived longer than about five years. With USB 3.1 we can say with a reasonable degree of confidence that low-cost SoCs will, some time over the next five to ten years, come out with USB 3.1, and in fact already there are RockChip SoCs that have USB 3.0. Once the USB 3.0 and 3.1 standards stabilise (which they yet haven't - in terms of both software *and* hardware) we will see more SoCs with it, and EOMA68 will be ready to package them up. > - Many first time hardware producers find that production takes longer > than they expect. Do you have any plans to avoid such problems? Do you > have definite manufacturers lined up? We do (or, in the case of mass-volume: we did). Working with them again will be hard, and it will definitely be important to go along to them with a fully-working sample as a demonstration unit. But: regarding production times, the entire EOMA68 strategy is *designed* to help mitigate production delays. Look at how the products are split into interoperable "base" units (we call them "Housings" now) which basically have nothing but simple peripherals (screen, keyboard, GPS, WIFI) and "Computer Cards". This scenario is covered in depth on the ecocomputing white paper [5] but the summary is that the housings (the base units) don't actually change - at all - and are in fact on a completely independent development lifecycle from the CPU Cards that go into them! So, a 15.6in EOMA68 Laptop Housing for example can be manufactured and sold indefinitely: for five to eight years if there's still demand for it, with the only reason for shutting down production (or modifying the design) being that one of the components goes end-of-life. It's a cash cow dream for manufacturers as well as suppliers! Imagine how happy a component supplier would be if he received a guaranteed order for a five to eight year span, in today's volatile market? Are you aware for example that there was a major recession in Guangdong when the Allwinner A10 came out? The previous incumbents were a Korean-based company that made a $13 800mhz ARM11 SoC. Based around this, tablet manufacturers scrambled to sell us $100 products. Within months, AMLogic came out with their first (GPL-violating) ARM Cortex processor, and sold it for $11. The Korean-based company's "Supernova" sales flash was over, that quick. AMLogic tablets sold for $50, so of course nobody bought the $100 ones. Then came along this little company called Allwinner, selling a 1ghz ARM processor that was $7.50 *including* the Power Management chip. They revealed a BOM of $15 for a tablet: actual products came out at $35 retail. Anyone left holding stock of the AMLogic components (bear in mind that the "Reference Design" basically gets copied verbatim, causing a huge demand not just for the SoC but also for associated components) basically had to go bankrupt, even though their clients would have put down cash up-front payments to buy the components. The price differential was so enormous that anybody who didn't have access to the Allwinner A10 was very quickly out of business, and that turned out to be a *lot* of people. The EOMA68 standard is designed *precisely* to mitigate against this kind of "Supernova Sales" effect. But not only that, because the Computer Cards (with whatever latest-and-greatest SoC is in favour today) can go into multiple products, there's plenty of room for factories to supply a huge demand across a *huge range* of products - not just tablets but also laptops, games consoles, hand-held camcorders, IPTV Set-Top boxes and many many more. And because of that shared usage across multiple products, the "Supernova" can burn even brighter, allowing the Foundries that produce the SoCs to dedicate their production lines for longer, thus increasing yields, thus bringing costs down due to more reliable and stable (but ultimately brief) demand. Compare this to the ever-frantic design cycle we face today, where SoCs end up in Single-Board designs where cost-cutting becomes really crucial, and we end up directly paying for that. As in, we *genuinely* end up paying more money for our electronics devices thanks to this desperately frantic cycle. > - Do you anticipate problems sourcing legacy PCMCIA cases? If a time > comes when you have to use something else, what would it be? This may surprise you, but we don't! PCMCIA is still used in the Satellite TV Industry, in the form of "Conditional Access Modules". That's the standard which was created so that whenever people broke the Satellite encryption, the manufacturers could send people a replacement hardware plugin instead of having to send them an entire new Satellite Receiver. Ironically, then, the whole modular approach is learned and used by the Satellite TV Industry, but in a grudging and unethical way. I'll be honest with you here, though (and realistic) - the Satellite TV Industry is on its way out. So we had better start taking over from them soon, before the tooling for PCMCIA is junked and the factory space made available for something else. If that goes, then we will definitely need to replace it, or find an investor or a MOQ 250,000 order. We'd be looking for an ethical investor who was happy not to have any say whatsoever in the EOMA68 standard. We learned that lesson already: EOMA68 cannot and will not be sacrificed for short-term profit, nor to suit any third party's special short-term product range. It's as simple as that: the standard *must* be inviolate, otherwise it cannot be trusted to *be* a standard. There are, incredibly, very very few well-designed standards out there. The only truly well-designed alternatives out there are COM-Express, M2.com, and the Arduino standard. Thats all! So in short: I'd rather keep to my integrity to do business in an ethical way. If EOMA68 cannot go ahead because PCMCIA becomes unavailable, and we cannot find an ethical investor, then EOMA68 dies. But, if you think about it: it's really not that hard to find a 1.27mm board-straddling dual-row socket and matching right-angled header, and to make up some 3D-printed casework and have some metal stamped out. If necessary, we'll do fine. I do have other standards that I could work on: EOMA54 re-uses CompactFlash, and EOMA200 is a PC-style standard that is similar to COM-Express, but EOMA68 is the one that I feel will have the largest potential impact, so we do that one first. > -Your hopes of getting a Respects Your Freedom certification suggests > that the BIOS will also be free. True? True - other than in the embedded world it's not called a "BIOS", it's called a bootloader (or bootstrap loader), which is often divided into several stages, each of which deals with the initialisation of various resources so that the next set may be accessed - it's not called bootstrapping without good reason! The Allwinner A20's bootloader procedure is documented on the Linux-sunxi wiki [8] and if you follow the trail you can find *all* the sources right from power-on all the way up to the linux kernel (and for Parabola ARM GNU/Linux-libre, even for the full OS). and, amazingly, find that they're all properly GPL (Copyright) compliant as well. Put it another way: We Don't Do Illegal Stuff And Claim It's Ethical, y'ken. > - The campaign page mentions recycling and bamboo and plywood > components. It looks like bamboo is used for the display casing. What > other uses are bamboo and plywood being used for? And can users get > replacements for the bamboo and plywood parts? Doll's houses! Radio-control models! Both of which are huge and stable specialist industries. Turns out that if you're in Europe and go on ebay you can find 1.5mm 3 ply birchwood that's a lot cheaper than bamboo plywood. There's actually a funny story about why I use bamboo for the panels. So, I'm at home, designing a laptop case on a 3D printer because the cost of doing injection-mold design requires $USD 250,000 minimum because the burrs created by the CNC machining need polishing for weeks at a time. I'm printing out parts as I design them, edges first, then I think, "hmmm, I need to fill the middle: I'm not going to waste my time and perfectly good plastic printing out a damn rectangle: what can I use instead that's err rectangular already? errr, I know: let's use plywood!" It wasn't until afterwards that I retrospectively added the whole "eco" thing on top, but it's pretty ironic that this was driven by my laziness and cost-cutting not wanting to waste time and plastic printing out a rectangle. And it wasn't until I'd naively ordered a set of bamboo panels from a specialist in the USA (to be shipped to Europe at great expense) that i found, six months later, that birchwood panels were 1/10th the price and a lot closer to home. At least I am making these mistakes and documenting them so that other people don't have to! > - It looks like the only difference between the two cards is the > distribution pre-installed. Is that true? If not, what are the other > differences? And why Parabola and Debian? It's correct, but please, it's important now that you've noticed this to emphasise the difference again, and it's a legitimate concern that was raised by the FSF. Bear in mind that whilst the FSF, as a non-profit foundation, is primarily focussed on privacy and freedom from spying and so on, from an ethical business perspective it's really much simpler: proprietary software costs you more money and a lot more stress. It's really that simple. Forget the fact that, as Benjamin Franklin famously said, "Those who would give up essential Liberty, to purchase a little temporary Safety, deserve neither Liberty nor Safety." [9]. I get quite a lot of people criticising the FSF for being "rigid" and "idealogical" as the rest of the world passes us by, but the "world" is rapidly turning into an Orwellian Surveillance State: Tim Cook only recently reminds us "We did it to ourselves" [10]. Now, the FSF has a particular nightmare scenario for the average end-user. They figure (rightly) that the average technical user (which includes anyone who can patiently and accurately and successfully follow instructions on how to install software) can look after themselves, and can discern the difference between "libre" and "not libre" software. It's everyone else that they're deeply concerned about. So with Debian having the synaptics package manager present end-users with an easy option to install "nonfree" software, with a simple check-box option but *no other warning as to the consequences*, that's the FSF's nightmare scenario. Bear in mind that whilst the majority of people view the FSF's advice as "irrelevant" and being as close to "religion" as you can possibly get in a technical world, the reality is that yes, as we've found out thanks to Edward Snowden and others, it's *not* ideology and it's definitely not conspiracy or paranoia when it's found to be genuinely Orwellian and we have backdoor processors in both major architectures (AMD and Intel) that are genuinely cause for concern [11] - enough certainly to cause China to pursue its own processor designs at enormous expense so that its Supercomputers and Government equipment remain free of spyware [12]. I always wonder whether companies that get visits from the NSA begin kicking themselves for not saying "No, you can't have the private keys" when their International sales plummet, but then we see evidence in some instances that they might *actually* have said "No" [13] So against this background, we felt that it was really really important to include a "Libre" Computer Card. In speaking with the FSF, they felt that it was important to have completely distinct names for the two products. The reason they said was that, just like in the drop-down example above, they did not want the average end-user to make the mistake of buying a Computer Card which they *thought* would respect their privacy but, because of an ordering error of a simple drop-down menu mistake that they overlooked, contained an OS that *didn't* warn them, but that nobody could actually tell them the difference without doing a thorough investigation. The discussions about why we included a non-free Operating System *at all* were particularly interesting. EOMA68 is an OS-agnostic hardware standard. There is even a "pass-through" concept on the TODO list [14] which, when plugged into an EOMA68 Laptop Housing for example, would allow the laptop's screen to be used as an HDMI Monitor, and its keyboard and trackpad as a standard keyboard and mouse on any other laptop or Desktop PC - and battery-operated at that. Also on the TODO list is an FPGA-based EOMA68 Computer Card. So I felt that it was important to get across, even at this early phase of the campaign, the general-purpose and open nature of the project. Think about it realistically: in the future there will be proprietary EOMA68 Computer Cards. Realistically, the quantities involved are likely to be larger, because that's "what people want". But along-side that will be the option for them to choose Libre (RYF Certified) Computer Cards, when they are ready. RIght now, thanks to the whole single-board design strategy which even involves hermetically sealing the case shut using ultrasound welding, the gap between the RYF World and the Orwellian World is so huge right now that they might as well be in separate Universes. But with the EOMA68 strategy, it's as simple as pressing a button and trying out the new Computer Card. So the *actual* choice of Operating Systems is not so relevant to me, but I recognise that for end-users it's important. In analysing choices of Operating Systems, however, Ubuntu has a proven track record of unethical decisions which tell us that they cannot be trusted to respect end-user privacy [15] or even Copyright Law [16], so it would be unethical to endorse their software by distributing and selling it. The best option then with the largest independent developer community behind it is clearly Debian, which, with its inviolate Social Charter provides us with a good basis to judge whether they can make responsible and ethical decisions. Parabola-ARM GNU/Linux-Libre was surprisingly hard to find. There are very few modern truly Libre OSes available, and there are even fewer modern ones that run on the ARM processor family. Parabola-ARM GNU/Linux-Libre was in fact literally *the* only available, modern and up-to-date OS. > - Assuming all goes well, what device will you develop next? (Aside: I > would really like to see a free tablet) Yeah me too, although honestly I've never been a huge fan of the whole tablet fad. Perhaps that comes from being a Software Engineer: I get totally frustrated trying to type at 150+ wpm with only one or two fingers. But, we'll get there. No, my favourite device that I really want to see is an EOMA68 Digital SLR Camera, where you could pop in an EOMA68 3G/4G/LTE capable Computer Card. To explain how powerful this combination would be: imagine in which alternate far-out universe it would be likely that either a Digital SLR Camera Company would have the in-house expertise to pull off a fully-functioning (even a proprietary) OS which *happened* to be running on a Camera, or that another Company which *did* have the in-house 3G / LTE expertise would remotely consider bringing in the enormous amounts of expertise necessary to bring out a Digital SLR Camera that happened to have a 3G/4G/LTE Modem in it? It's not going to happen, is it? And both of them would freak out massively at the the slightest thought or suggestion by any of their respective Engineering Teams that they should consider doing it as a Software Libre project! Now let's separate the two. An EOMA68-based Digital SLR Camera with some form of proprietary EOMA68 Computer Card, with a locked-down OS based around the open EOMA68 standard? > - Is there anything else you would like to say? Yes there is. You can see from that very first work that I did with Samba (and later, what many people are not aware of is that I did the same thing with Exchange 5.5), is that at a critical point in Free Software's relevance, well before companies like Canonical and Google were household names, I helped bridge the yawning gap between the proprietary Windows and the free software worlds. I broke the ice, and didn't get murdered for doing so (this is not said as a joke), so other people felt safe to do the same thing. Proprietary corporations are now outright taking blatant advantage of Free Software and are in danger of doing the exact same thing - creating polarisation - but this time in hardware form: Mediatek (GPL violations), Apple (BSD-based and not releasing it back), Google (Android as BSD-based). LG actually took legal advice on how to do Tivoisation [17], and they actually consider it to be a *failure* if you even *notice* their rampant GPL violations in any of their products. A similar polarisation and monoculturisation is even occurring within the Free Software world, with the dangerous introduction and accidental standardisation around systemd [18]. This project is therefore at its heart an ethical initiative to give us our freedom of choice back, at the hardware level. It's damn ambitious, but I am at heart a logical person that is unaffected by the scope or scale of the task that I can clearly see has to begin somewhere, with someone taking the first step. Been there, done that, still alive, so let's try another step. The eco-benefits are ironically an incidental side-effect of the reduced long-term cost of ownership of EOMA68 devices. There isn't anyone else in the world that is really thinking about these issues, so it actually becomes incredibly challenging to put them across in a simplistic way. The money-saving aspect is the best one that we could come up with. It helps that, long-term, it's true. But to get there, our campaign needs support. We've received a lot of flak for designing what appears to be "mediocre" hardware, but that's what can be achieved on a small budget. Imagine what I could achieve if you supported this campaign *now* so that it took off and I could expand it? > (Another aside: any thoughts about the pi-top, which does some -- but > not all -- of what you are planning) No. It does absolutely none of them, and as such it provides a very good counterpoint. * The casework for the pi-top was 3D printed as a single piece that took them 50 hours of round-the-clock monitoring to make sure that the printer completed it: the EOMA68 Laptop housing is broken down into 35 separate parts so that any one of them, if it breaks (during use or during printing) does not mean that you have to throw the entire casework away. * The PCB schematics for the pi-top were *NOT* made available until after the campaign had ended. By contrast the only designs being kept back are the EOMA68 Computer Cards so that we do not get copies (in China) jeapordising the project at this very early phase. * The development and progress of the pi-top was NOT made public. There was no consultation or invitation to participate extended to either the Software Libre or Hardware communities. * The casework of the pi-top has entirely been converted to injection-molding, and the CAD files were not made available until after the product was shipped. The 15.6in laptop casework designs will *always* be available, *have* always been available as a GPLv3+ licensed software project right from the start [19]. * The overall thickness of the pi-top is enormous, because it has to house a 15mm high PCB. EOMA68 Computer Cards are only 5mm thick, so the 15.6in Laptop is only something like 21mm high in total. * The pi-top is a small niche market novelty laptop which can expect to have a limited lifespan, whereas EOMA68 is targetted specifically at the world-wide mass-volume computing appliances market. Overall, then, if you've used a pi-top, you'll know that it doesn't "gel". It's completely different in other words, and I'm grateful to you for mentioning because by counterexample it sends a clear message to people that I genuinely do follow ethical business practices, not pay them lip-service because it might unwittingly attract buyers because the word "open" happened to be used in the marketing material. All the mistakes that I've made are out there for other people to learn from, just as I learned from the mistakes made by other Open projects such as the OpenMoko, OpenPandora and many others. > ---- > And good luck! I am seriously considering ordering a kit myself. thanks, that's really appreciated. [1] http://lkcl.net/articles/welcome.to.samba.domain.html [2] https://bugs.winehq.org/show_bug.cgi?id=17195 [3] https://blogs.fsfe.org/pboddie/?p=1305#comments [4] http://elinux.org/Embedded_Open_Modular_Architecture/EOMA-68 [5] http://rhombus-tech.net/whitepapers/ecocomputing_07sep2015/ [6] https://mjg59.dreamwidth.org/8991.html [7] http://www.codon.org.uk/~mjg59/android_tablets/ [8] http://linux-sunxi.org/BROM [9] https://techcrunch.com/2014/02/14/how-the-world-butchered-benjamin-franklins-quote-on-liberty-vs-security/ [10] http://time.com/4262480/tim-cook-apple-fbi-2/ [11] https://slashdot.org/comments.pl?sid=9257279&cid=52339563 [12] http://www.theverge.com/2016/6/20/11975356/chinese-supercomputer-worlds-fastes-taihulight [13] http://www.infoworld.com/article/2608141/internet-privacy/snowden--the-nsa-planted-backdoors-in-cisco-products.html [14] http://elinux.org/Embedded_Open_Modular_Architecture/EOMA-68/Passthrough [15] https://www.eff.org/deeplinks/2012/10/privacy-ubuntu-1210-amazon-ads-and-data-leaks [16] https://sfconservancy.org/blog/2016/feb/25/zfs-and-linux/ [17] https://en.wikipedia.org/wiki/Tivoization [18] https://linux.slashdot.org/comments.pl?sid=8387451&cid=51004979 [19] https://www.youmagine.com/designs/libre-hardware-licensed-parametric-laptop-design