Starting an Open Access Journal: a step-by-step guide part 1
Introduction
Scholarly publishing is totally broken. Not only, at present, can most of the people (taxpayers) who fund research not get access to it, but plans to fix this look set to screw over Early Career Researchers and anybody else who can't persuade their funders to give them the up-front fees required by publishers for Open Access journals.

There are other models. I have proposed that the university library could function as a re-invented university press. However, this guide is intended, over the course of as many parts as I need to be able to write this in manageable chunks, to signpost a third way. This guide is for academics who want to establish their own journals that are:

Peer reviewed, in a traditional pre-review model
Open Access and free in monetary terms for authors and readers
Preserved, safe and archived in the event of catastrophe or fold
Reputable: run by consensus of leaders in a field

Pre-requisites
My assumptions about you:

You are a non-commercial publisher, individual or organisation with revenues under $25,000
You are working in the humanities. If you are a scientist, this guide may still be of use, but there may be aspects of your field that I overlook; you'll have to fill in those gaps yourself
I'll explain each of these items as we go through, but the monetary costs, on a yearly basis are:

Web hosting: $60/year (approx. For example: Bluehost) (NB. this is an affiliate link, for which I will receive commission. There are many other hosts out there if you want to search on your own.)
DOI numbers/CrossRef membership: $275/year. Alternatively, you can join the OASPA and get CrossRef membership included in their fee of €75/year (including 50 DOI numbers)
CLOCKSS archival service: $200/year
As you can see, the yearly financial cost is approximately $350/year, which is equivalent to £225/year. If there are several of you within an umbrella organization (again, it would have to be non-profit), you could share the DOI prefix, CLOCKSS membership and hosting, thus making an economy of scale possible.

Social vs. Technical
With enough persistence, I'd argue, anybody with a mild technical competence and enough persistence could install Open Journal Systems, the software that I'm going to be using in this guide. What takes the time, energy and willpower is to get the social, rather thantechnical aspects sorted. What do I mean by this?

Editorial board
Peer reviewers
Copy editors
Proofreaders
Editorial board
The board is absolutely crucial. Academic journals work on a system of academic capital; you need respected individuals who are willing to sit on your board, even if they are only lending their name and you end up doing most of the legwork. It should only be a matter of time before academics realise that journal brand isn't (or shouldn't be) affiliated to publishers, but rather to the academics who choose to endow a journal with their support. Get good people who are respected within your discipline(s) and you're on the right track. Here's an example email that I used to ask people to sit on the board for Orbit:

Dear X,

I don't know if you remember me or not, but we were both at the ___________ conference last year and it was in relation to this that your name came up. We were all very impressed by your paper and wondered if the following proposal might be of interest to you.

I have been working over the past few months to put together a new journal of _________ scholarship. I now have (finalising this week) the requisite funding. The website is done, complete with online submission and peer review systems. The editorial board thus far consists of myself, _______, ________ and ________. _________ has expressed tentative interest and we are also contacting _________, __________ and __________.

Anyway, I was wondering whether you'd be interested in being either on the editorial board itself or acting as a peer reviewer? If you had any work you'd be interested in contributing, we'd also be interested. We have five specific aims, one of which is to get all articles that are accepted published within 5 months. In fact, I've attached a statement of purpose to this email; it's confidential, so please don't circulate, but it might be of interest. The mockup title page harks back to before we had a better name...

If you wanted to take a look at the site:

http://www.myjournaladdress.com

Temporary username: a_username
Temporary password: a_password

You won't be able to see very much, as all the editorial functions require a proper username and password (obviously!) and there aren't any articles yet, but it will give you a feel.

The tentative name for the journal is: "_____________".

Let me know what you think and whether you're interested; I personally think this is incredibly exciting and am itching to go!

In the meantime, hope you're well and best wishes,

Martin
You may also want to incorporate some statistics on library spending and stress the open nature of the journal, but you get the idea.

Peer reviewers
When the first articles start flooding in, you'll need all the help you can get. These have to be people you can trust to understand the challenges you're facing. They need to set the bar high for the first issue while also appreciating the difficulties of attracting the big names to start-up journals. Contact people early so that you're ready to go.

Copy editors and proof readers
If you end up doing this job yourself, expect to spend about 16 hours per article, including typesetting (depending on how you do this -- covered later). Therefore, get some good people to help you. They should be sensitive to individual style (ie. light-touch editors) but who also will bring some sort of synthesis to the style.

Other decisions
You need to make some decisions and take action on the following issues, not all of which I can describe for you:

Journal name(!), scope and remit
OA policy (I'd recommend Creative Commons Attribution) and copyright stance (let your authors keep their copyright)
Publishing mode (issues or rolling? Do issues always make sense in an online environment, or should you just publish as submissions arrive?)
Initial CFP
Timing (don't time it so that all your first submissions arrive in the Christmas break, when nobody can review them, for example)
Anyway, that's all for Part 1 (I have to do some work!), but in part 2I'll start to cover setting up the technical side of things and the timescales for applying to various organizations.