The future of JuliaBox


#1

I heard on the main Gitter that JuliaBox is set to go commercial at the end of the year.

I don’t think that (as it stands) this strategy would work in the best interests of the creators.

However I think that the following modification would work very well:

Continue to make the basic service free, but CPU cycles per second and cycles per week are capped on the free version. One can purchase computing power (a bit like Amazon EC2).

I think it’s important that the basic JuliaBox remains free. Because academics have for many years used Matlab code to support academic papers. And it is a bad situation as Matlab is a commercial product (and an expensive one at that).

It is in everyone’s interests (apart from Matlab’s) that academia switch to a free available alternative platform. In the new-age individuals are able to contribute to scientific progress without any University affiliation.

Julia looks set to step in and take the reins.

If JuliaBox it Is kept free and builds the functionality to exchange Julia notebooks via a http://juliabox.com/myNotebook style URL, I’m sure Academic papers will start using it as a supplement (maybe even as the main paper one day!). And that will advertise it. The more people that use it, the more people will pay for extra compute-power.

I personally love the idea of developing an algorithm on my home machine, loading it up on JuliaBox, loading up 50 bucks of compute power with PayPal and picking up the results the next morning.

You could even tier the compute-power, so top tier would claim priority, lower tier would wait until low load. A bit like booking seats on planes: look further in advance and it’s cheaper.


#2

I don’t know anything about the plans for JuliaBox, but it is important to realize that Julia will remain free/open-source software regardless of what happens to JuliaBox. Right now, in my experience most people only use JuliaBox when they are first trying things out with Julia, e.g. students who are trying to do a homework problem and don’t want to worry about installation, and they quickly switch to installing Julia on their own machines.


#3

I’m a little confused by your post:

I personally love the idea of developing an algorithm on my home machine, loading it up on JuliaBox, loading up 50 bucks of compute power with PayPal and picking up the results the next morning.

Isn’t that exactly a paid computing service? It would be great to keep JuliaBox entirely free, but the service has so far been running on very generous grants from cloud computing services. Those grants will run out at some point, after which the service needs to be partly paid or cease to exist. Ideally some level of service will remain free since it’s a great tool for people to try Julia out.


#4

Sorry, I wasn’t making my point well.

Yes, I’m saying I would be happy to pay for that service! i.e. I’m not trying to argue for a free lunch.

But at the same time I would also hope a free basic service emerges which allows easy sharing (and maybe even collaboration) of Julia projects.

The point I wanted to make was that a combination of the free service and a paid service may make good business sense; the free service could become the de-facto standard for sharing code in Academia, which would advertise the paid service better than anything else.

As an aside, when I was working with C++ http://ideone.com/ was hugely valuable. Engineers in ##c++ (on IRC Freenode) were using it constantly for problems and solutions. You can clone someone else’s ‘paste’ in a single click, run it (remotely), modify it, send back a new URL. It worked very well.


#5

I will second that. The plan is to keep JuliaBox at some basic level completely free to make it easy for newcomers to try it out - so long as we keep finding grants. However, we do want to add things like easy parallel computing, extra storage, and those will need to be charged for.