It's 7:30 on a Wednesday morning. I've been up for two hours, there's no hot water, a leak in the roof and I really should drink some coffee. Yet I've finally come to the decision I knew I was going to take. I'm shelving Atomic Droplet. And I both hate it, yet feel an overpowering sense of relief.

Atomic Droplet came into being after talking to ... let's say a prominent funder of good ideas ... once the ruckus from Virtual Katy died down in late 2006. The exact words spoken were "if you present us with a good idea, we will fund it", followed by a close definition of what a good idea actually was. OK, said I, and went off and did some research to find an opportunity that fulfilled this criteria. I found that (a) there are loads of "b-league" advertising networks making very real money and that (b) there are many billions of closely targetted impressions going begging in web forums. Part (b), BTW, was established by writing a massive screen scraper in python that spent a few days looking at the number of impressions on a per-thread basis over 350 forums. Obviously I have the code in svn somewhere if anyone's interested.

This was then added to some industry figures and transformed into a short and rough business plan, hitting the provided definitions of "good idea", and that on skim reading it now still appears to a well reasoned and quite sensible piece of writing. Hmmm. Said plan was then taken back to said funders who appear to have changed staff with the new people being somewhat less keen, although not exactly distant. I said I'd get it up and going in a rough form, then get back to them. This was probably a mistake.

My next mistake was involving FRST. I figured that since I would be doing R&D, that FRST would be able to help. The application took a surprisingly precious few weeks and the conversation went like this: yes, yes, yes, yes, yes, yes, no. This, clearly, was also a mistake and it led to my next major problem...

VoiceQ were on a push to get their software into a very big name client and really needed my help to make it happen. They've been my clients for a long time, have invested heavily in their business and I really do feel a resposibility towards them so I agreed to work with them to make this happen for a few months. This may be classed as a mistake but I don't feel that ethically I really had a choice other than to be not a vital part of their organisation in the first place. After this I returned to Atomic Droplet work but, by now, we were getting close to a *year* after the initial work had been done.

Miraculously I then did something right. I looked at the work to date, hated it, and went back to the drawing board significantly - wireframing the site with OmniGraffle and using the services of Chris Beaven to do CSS and, eventually, a little Django work. Much of the back end code could be reused (and was) and the project, basically, started to come together really well. I also got an arrangement to be able to use Bassdrop as a beta site. It got 80% done, and we all know how long the other 20% takes.

I then got an offer of a short contract with shinywhitebox, starting in the new year. The idea was to do the remaining 20% in evenings and weekends while I did the shinywhitebox gig, supposed to last three months, with the site actually being up and running by the time swb gig ended. Things did not go entirely according to plan. The swb contract extended out to more like eight or nine months, finally coming to an end a few weeks after Burning Man in September. Again, doing this contract could be classed as a mistake but I want to make something clear - I did some great work for swb and am thrilled to have been a part of making a piece of software that's well written, useful for a large number of people, and that will make my client a great deal of money. I also got an insight into the market for niche (but consumer) mac applications, learned some new techniques, and made a good friend, colleague and ally in Neil.

So, broadly, this takes me to today. Atomic Droplet has actually been up, running and serving advertisements in a trial fashion for about six months. It's tested (the ad server has been through a few iterations); has video tutorials; can charge money and distribute it to the appropriate recipients (including the tax man) and generally speaking works. So we have to ask: why not press the big red button, declare it to be "working" and go make money? Well:

* The business was always designed as something that would need investment, but would scale with that investment. It has been, realistically, far too long to go back to the original funders now. I can't realistically scale it with my own funds or from revenues so that would more or less doom it to being a small business which I don't realistically feel I can spend the time on. Especially since...

* Once you press the big red button and start taking people's money, integrating into forums etc. etc. you have to continue to support clients at both ends. If the business is not going to work the whole thing becomes a massive time sink - "selling problems" as a software distributor once called it. Don't want to sell problems, especially if I'm not going to make much money doing it. Which leads on to...

* One of the critical elements of scaling was being able to tie into other advertising networks. Two things about this: one is that the demand/supply curve is very much tilted towards there being too much supply of advertising space and too little demand for it. One marketplace set me up with a VP in LA until they realised that I was looking for more banners rather than more space - at which point I was shunted off to a web form and forgotten about. Secondly a lot of these marketplaces deal only with US specific requirements. With time, funding and scale we may well have got over this, but the smart solution would be to change to NZ specific advertisers...

* I tried to form an alliance with a group looking to do NZ specific advertising. This group were executive heavy, technology light and an offer was made to modify Droplet to their needs and treat it as an equity investment. I.e. with no money exchanging hands. However, talks appear to have stalled with the CEO merely failing to respond to emails - an disgraceful way to treat a potential investor.

* I also treated small business and large business marketers as having more or less the same requirements merely acting at different scales. I don't, now, believe this to be true.

In short it seemed there was a great possibility that I was about to be throwing good time after bad and entering into worse commitments, particularly without solid funding to enable growth into the higher margin campaigns.

Sigh. This has taken all morning, what with screenshots and dealing with the plumbers so I'll attempt to draw a line under it now. Despite all my bitching and whining, the actual resulting Atomic Droplet webapp is an awesome piece of work. I'm particularly proud of the ability to build an advertising campaign from start to finish - including uploading assets, bidding on advertising space and costing - from a single page. Apart from that it just works. Pages display the information you need, clearly. The UI adheres to the golden rule of not specifying in what order operations must be performed (you listening, web interaction designers? didn't think so). New campaigns are provisioned onto the advertising servers instantly. It is a much much cleaner interface than any of the other advertising networks I surveyed either before starting the project or since. As a first Web2 application I pretty well hit it out the ballpark but ... well .... what can I say? ... dammit.

There is something else coming, of course, but I have already tried the reader's patience. In the meanwhile I can only leave you with a few short screencasts:

* A quick video I made of trundling round the UI this morning prior to the big shutdown (or download as a better quality quicktime).

* The mondo embarrassing "introduction" screencast

* And how to do your first and subsequent campaigns on the UI.

Sigh. Sadness. 'Tis all in the nature of entrepreneurial exploits I am afraid.