Megapixels and libcamera were happening without Purism, so I agree on that point. However, NXP poorly documented the CSI interface on its i.MX 8M chips and other projects like the MNT Reform don’t need it, so I doubt that anybody would have bothered to figure out what Purism did with the L5’s cameras, since it took months of painful trial and error. For example, one guy posted on the NXP forum:
I’m not entirely sure where to go for documentation as well, because it seems like the iMX8M Reference Manual has almost zero information on the CSI interface, and there are only 1-2 replies in forums saying that this device is somewhat of a mix of the iMX7 and iMX6 interfaces, regarding MIPI and CSI (?!)
If Purism had dropped development of the L5 in Feb. 2020 (when they decided to stop giving refunds), there would have been so few people who saw Purism’s code working that I doubt that any volunteers would have arisen to carry on the code projects that Purism started. By early 2020, only the few hundred people who got the L5 dev kits, Birch and Chestnut had seen libhandy and Phosh in action. Remember that the Mobian distro with Phosh started in May 2020, Debian first added Phosh in June 2020, the Pinephone postmarketOS Community Edition with Phosh was announced in June 2020, Fedora added Phosh in July 2020, and GNOME adopted libhandy as an official project in August 2020.
Furthermore, the history of GNOME shows that it hasn’t incorporated mobile code in the past if that code was abandoned by the developer. GNOME announced the GNOME Mobile & Embedded Initative in April 2007, and a month later Nokia offered the GNOME Foundation its Hildon code from Maemo to be incorporated over time into GTK. Maemo’s hildon-gtk made GTK 2 responsive and touch friendly, in a similar way that libhandy did for GTK 3. GNOME officially adopted Hildon as a project, yet GNOME relied on Nokia to do the work to incorporate it into GTK. Nokia bought Trolltech in January 2008 and switched to Qt as its mobile toolkit, so development of Hildon was abandoned and it wasn’t incorporated into GTK. The GNOME developers largely ignored the ideas in Hildon when designing GTK 3 (although they did support touch screens).
Despite a great deal of interest from GNOME users over the years, the GNOME Foundation did little to make its code mobile friendly, which I attribute to the lack of interest from GNOME’s major contributors. Red Hat and SUSE make most of their money from software and services running on servers. Google had no interest in developing GNOME Mobile as a competitor to Android and Canonical was developing its own mobile interface based on Qt. It was Purism’s work on libhandy/libadwaita and Purism paying Tobias Bernard to work on the design of a mobile interface that led to Bernard and Jonas Dreßler applying for a grant from the Prototype Fund for mobile development in GNOME 40+, which kickstarted GNOME Shell Mobile in April 2022
Given that GNOME made little progress on its mobile initiative between 2007 and 2020 without Purism, I see little reason to believe that GTK/GNOME volunteers would have picked up libhandy if Purism had abandoned it. The essential difference between Nokia’s Hildon and Purism’s libhandy/libadwaita was the fact that Purism worked actively with the GTK/GNOME projects to incorporate their code, and that work started in the second half of 2020 after Purism was already having financial trouble.
Yes, I am biased, but that doesn’t change the fact that most people who preordered the L5 would be in a worse situation financially if Purism had declared bankruptcy in early 2020. Only those who cancelled their orders before Purism ran out of funds would have benefited, whereas the roughly 5400 people* who have received the Librem 5 and a sizable percentage of the 600 people who are currently waiting for their refunds would have been harmed financially. In contrast, today those 600 can either accept the Librem 5 and resell it or continue to wait for Purism to get enough new orders to be able to repay them. Either way, those people are probably better off financially with those two options than in a bankruptcy proceedings.
My position is to pressure Purism to make a public commitment to pay out the refund requests with the funds that it receives from new orders of other products, as Purism stated in its private letter to Rossman at 7:00:
The production of the phone ended up costing more than people had prepaid to order it. Another paradox that I discovered was that it was cheaper for Purism to refund the phone than to send the phone–the difference was financed through the sale of other products. Todd decided to honor his commitment to the majority and keep his investment in the R&D, while honoring the refund requests over time by funding them with other products.
Telling people to not buy Purism’s products makes it more likely that Purism won’t have new orders to pay the refunds, so I don’t understand what is your goal. If your goal is to prevent new people from being harmed, that doesn’t make much sense because people ordering from Purism today are not taking a risk like they did in preordering the L5, since the only new product for sale is the Librem 11 which Purism claims will ship within 10 days of ordering. If your goal is to punish Purism for its bad behavior, then continue publicly denouncing the company, but I don’t think that is going to help people get their refunds faster.
I can see the argument that Purism should be punished as a matter of principal, because it violated the rule of law and undermined public trust in the system of commerce. However, if we are going to look at this from the perspective of how many people were harmed financially, bankruptcy probably would not have resulted in a good outcome for most people who cancelled their orders.
As for your argument about a “fair order” in bankruptcy, I would say that customers who ordered a good or service from a company should have higher priority in getting paid than secured creditors (i.e. banks that loaned to the company), but that is not the way that the law works.
It is also worth pointing out that many of the people who refuse to accept the Librem 5 and resell it are not doing this for economic reasons, since most of them paid $599-$799 for a product that now has a list price of $999.
First of all, we should be grateful that anyone is willing to invest money in Purism considering how many business failures there have been with mobile Linux in the past. Investing in the development of mobile Linux is hardly a rational business decision considering that history.
The business owners and investors in Purism have taken a loss developing the L5, according to Purism’s letter to Rossmann. Given how tarnished Purism’s reputation is from developing the L5 and all the bad reviews it has received, Purism is unlikely to ever recoup what it invested in the project since it isn’t going to sell many L5’s in the future.
Purism reported that it received $2.81 million in preorders for the L5 by 2018-11-01, so let’s guesstimate $4 million total in preorders (since the preorders slowed down a lot over the course of 2018, and $4M would be 6000 preorders at an average price of $667, which aligns with the numbers in Purism’s letter to Rossmann*).
If we assume that Purism pays developers $5k per month, then I calculate that Purism has spent around $2.2 million developing the software in the L5.
| Purism employee | Title | Work | Started | Stopped | Months |
|---|---|---|---|---|---|
| Guido Gunther | Distinguished Engineer | phosh, phoc, libhandy, feedbackd, gtherm, linux-next | 2018-05-07 | 65.4 | |
| Mohammed Sadiq | GTK Developer | Chatty, gtk | 2019-06-04 | 52.3 | |
| Adrien Plazas (purism) | GTK Developer | libhandy, geary, epiphany, libadwaita, pureos-store | 2018-02-01 | 68.6 | |
| Sebastian Krzyszkowiak | Shell Developer | mobile apps, phosh, millipixels | 2019-07-03 | 51.4 | |
| Dorota Czaplejewicz | Application Developer | Squeekboard, camera configuration, Millipixels | 2017-12-08 | 70.4 | |
| Julian Sparber (½ time) | Application Developer | Calls, Fractal, Phosh | 2019-06-23 | 2023-04-27 | 23.4 |
| Alice Mikhaylenko | Application Developer | libhandy, epiphany, kgx | 2020-09-22 | 18.2 | |
| Tobias Bernard (1/3 time) | Lead UI/UX Designer | design of Phosh and future camera app | 2018-05-30 | 2021-11-22 | 14.1 |
| Andrea Schaefer | Application Developer | chatty | 2018-05-28 | 2020-07-20 | 26.1 |
| Bob Ham | Phone Developer | Calls, ModemManager, haegtesse | 2018-02-01 | 2020-03-05 | 25.4 |
| Sam Hewitt (1/3 time) | UI/UX Designer | Design of chatty & millipixels | 2019-07-30 | 16.8 | |
| Martin Kepplinger | Kernel Developer | linux, | 2019-05-31 | 26.2 | |
| Evangelos Ribeiro Tzaras | Application Developer | GTK4, libadwaita, chatty, calls | 2020-12-03 | 17.0 | |
| Angus Ainslie | Full Stack Developer | uboot, agps, zephyr, linux | 2018-05-07 | 32.7 | |
| Total months | 432.3 | ||||
| Total cost (assume $5k/mth) | $2168k |
If we estimate the bill of materials was $400 per phone, that would cost $2.16 million for 5400 phones. Then add in the cost of designing the phone, which includes paying some design firm in Taiwan or China and the salaries of Nicole Færber and Eric Kuzmenko. Faerber reported that the main circuit board went through over a dozen iterations. With some back of the envelope math, it isn’t hard to see how Purism spent over $4M on the L5. Canonical estimated that it would have needed $32 million in preorders to pay for its Ubuntu Edge phone, so its hardly surprising that Purism ran into financial trouble.
However, there is a larger argument that Purism’s dev work benefits society as a whole, which should be taken into account because sticking it to the rich Purism investors means harming a lot of other people. For example, the Pine64 survey in Jan 2022 found that roughly 35% of PinePhone owners are using Phosh. Given that the 3079 respondents to the poll represented “less than 5%” of the PinePhone owners, that works out to over 60,000 PinePhone sales in 2 years, or about 100,000 by now, so roughly 35,000 PinePhone users are benefiting from Purism dev work. In addition, there are roughly 30 million desktop Linux users in the world, and the majority are using GTK/GNOME software that has been improved by libhandy/libadwaita.
Then consider the long-term goal for mobile Linux to become a viable alternative to Android and iOS, so people have the option to not be controlled and manipulated by powerful companies and big moneyed interests. People should control the technology in their pockets rather than it becoming a means to spy on their personal lives and monetize them through targeted advertising.
When big tech companies design cell phones to only last a couple years, they are pushing millions of poor people to keep buying expensive planned-obsolescent devices. To get to a world where cell phones can last 10 years rather than 3 years, we need mobile Linux to succeed and that will only happen if companies like Purism are paying developers to work on it, rather than driving those companies into bankruptcy.
As I see it, the goal is for mobile Linux to get good enough that it goes beyond the Linux enthusiast market and starts reaching normal people. At some point there will be enough demand to convince the makers of the leading-edge mobile SoC’s (Qualcomm, MediaTek, Samsung, UNISOC and HiSilicon) to start supporting Linux, rather than just supporting Android. If the bootloader isn’t locked and the SoC is fully supported by Linux, then people can switch their phones from Android to Linux to keep using them for many more years and avoid the harvesting of their personal data. Extending the life of mobile devices with Linux will dramatically lower their economic and environmental costs.
I see little chance of mobile Linux succeeding in a reasonable time frame without paid developers working on it, and I see little hope for AOSP derivatives ever reaching beyond a niche market, because Google can simply stop releasing new AOSP versions if it ever becomes a true threat to Google’s profits. It is for these reasons, why I advocate for people to support Purism despite its flaws, because I don’t see another company that is willing to invest in mobile Linux in the right way. Unlike Palm/HP/LG’s WebOS, Canonical’s Ubuntu Touch, Mozilla’s FirefoxOS and Samsung’s Tizen, Purism has tried to upstream its code as much as possible, and get its code incorporating into the principal Linux distros, rather than building unsustainable code silos, which is why the majority of the 344k lines of code in projects that Purism started are now part of GNOME.
* Purism’s letter to Rossman said that 10% cancelled their L5 orders and there are 600 outstanding refund requests, so we can calculate 5400 orders that weren’t cancelled, but that ignores the number of refunds that have already been paid, so the number of uncancelled orders might be larger.