Re: The future of mobile computing.

[Olaf Buddenhagen]

Hi,

On Wed, Dec 10, 2014 at 03:52:40PM -0800, Omar Radwan wrote:

> >Are you familiar with the history of
> >http://en.wikipedia.org/wiki/Openmoko ?
[...]
> I am familiar with it, and how it didn't spark enough interest,

I wouldn't say it didn't spark enough interest... The problem was more
that many of those people who were interested in it and sitting on the
fence, ultimately refrained from buying it because they saw the reports
from early adopters, of both the hardware and the software being
unfished to the point of being virtually unusable.

There are many things that went wrong in the OpenMoko project. They
considerably underestimated the time/effort to create a viable software
stack. They considerably underestimated the time/effort for hardware
field testing and refinement to make it reliable. They underestimated
the amount of QA necessary to gain consistent quality in mass
production. They underestimated the time necessary for establishing the
product(s) to a point where sales would keep it sustainable. In the end
it all boilds down to the conclusion that in spite of the massive
investment (several millions I am sure), the project was still severely
understaffed and underfunded for what they attempted to do.

In hindsight it appears that it might have worked better if certain
things were done differently: using an off-the-shelf modem module
instead of a custom design (more expensive per piece, but *way* less
headaches); creating a device with a keyboard, which most of the
potential customers at the time would have preferred; sticking to one
software stack -- in spite of some perceived shortcomings -- rather than
starting over several times; not marketing the product as ready for end
users long before it actually would have been... But then, all of these
would have implied other problems -- so we never really know. We can't
assume that a new project would make fewer mistakes. It's part of the
risk, and we have to calculate with that.

Note that today it would be somewhat easier to do something similar: we
have a pretty good idea nowadays what people expect from mobile hardware
and software; we have several existing software stacks, which would
"only" need adaptation for the device, and possibly some polish; I think
some of the hardware components needed are more of a commodity nowadays;
and I think hardware prototyping is somethat easier too. Still, it would
be a very expensive and risky project -- while not impossible to
succeed, I fear the FSF can't afford to run such an expensive experiment
:-(

(Also note that this wasn't by far the only open hardware project that
failed... Just the most prominent one I guess.)

> Jolla is a multi million dollar company after just 3 years.

Jolla was a multi million dollar company to begin with... More on this
below.

> We just need a hardware design, manufacturers that will agree to
> assemble the hardware and either give us Free firmware, or make it
> easy for us to write our own, and start an indiegogo campaign. To get
> to the indiegogo campaign it shouldn't take us more than a year. And I
> do believe that alot of people will want the phone by then, if we
> generate enough hype. But not like Ubuntu and promise something for 3
> years that still hasn't come out. We have to set an exact date and
> make sure the phone is released by that date.

And just how do you expect to get that hardware design -- and in such a
timeframe, too?

Just to give you some perspective: established mobile phone vendors --
with streamlined processes, and *huge*, experienced developer teams --
need two years to bring a new model to market.

Did it occur to you that there are *reasons* for the Ubuntu delays?...

Also note that crowdfunding, while it allows bringing hardware to market
with less initial capital, carries some serious drawbacks too. Most
notably, only a fraction of the potential buyers will participate:
others will either not hear about it at all, or won't want to commit so
quickly. That means a smaller initial production run, and thus
considerably less affordable devices. (Economy of scale is a *huge*
issue here.) Which in turn means even fewer potential buyers will want
to afford it...

All that means that the initial reach is much smaller, and thus
considerably less helping hands to polish the software. (Especially as
those who can afford such an expensive device tend to have little spare
time for hacking...) That further contributes to slowing down the growth
of the customer base, etc. Making it even less likely that the device
reaches a meaningful number of users before becoming obsolete...

On Thu, Dec 11, 2014 at 07:12:20AM -0800, Omar Radwan wrote:

> We can manage it if Jolla managed it, they had less funds, more work
> to do, and probably alot more difficulty.

You seem to have some very mistaken ideas about Jolla...

This wasn't a bunch of enthusiasts creating a phone in a garage with
their pocket money as capital, or any such thing. It's a major
enterprise, set up by experienced Nokia managers and engineers, with
some serious venture capital.

Moreover, their success so far is quite debatable. They essentially have
a niche product, at a rather high price for medicore specs and
half-baked software, interesting only to ardent Nokia-lovers in exile,
willing to pay a large premium for sentimental value alone. This niche
is not sustainable though in the long run: it can only get smaller as
more and more people get over the loss of Nokia and move on. (And I have
some serious doubts it's sustainable even now -- I suspect the sales
barely cover the production costs, and they are probably using up
capital to keep their ongoing operations running...)

With what you say about their licensing policy, they are obviously not
seriously aiming at free software enthusiasts either. (A smaller, but
more sustainable niche...) Rather, they seem so convinced of the
superiority of their OS, that they believe it will be sufficient as a
unique selling point in the mass market. (And needs to be "protected"
from competitors...) However, from what I have seen, that doesn't seem
match outside perception -- so unless their design is so much superior
that they can soar past the competitors in spite of limited development
resources (highly doubtful), they are pretty much set up for failure...

Finally, I think you are overestimating the ambitiousness of their
hardware design. While this is mere speculation, form what I know I
suspect it's just some slightly modified reference design from the
actual manufacturer. For the lowest system layer with kernel and
drivers, they are using an Android environment (probably also based on a
reference image from the manufacturer...); with an adapter library on
top so they can run a more or less "normal" GNU/Linux-based system with
Wayland etc. They probably couldn't release driver source if they wanted
to -- because they don't have it in the first place; just using standard
drivers from the component vendors and/or device manufacturer.

-antrik-