Agile, waterfall and muppets

There’s been some very good debate of late about how to do it all better, with a heavy emphasis on the role that Agile methods might play. “It” in this case being not just government technology, but extending to policy development, communications and more.

There seems to be a massed rebellion against the substandard, the lame, the failed and the apparent deathgrip that a few large suppliers have on taxpayer billions. This is a very good thing, of course.

I found a couple of recent pieces very insightful (in different ways). Alistair Maughan kicked off a lot of debate with a provocative piece arguing that Agile was doomed to fail in a public service setting. This from one of the architects of arguably some of the biggest, most expensive and (inarguably) rigid ICT contracts imaginable. Adam McGreggor from Rewired State came back with evidence of Agile success, and a rebuttal of the good lawyer’s key arguments.

These arguments really seem to me to be based around an overarching sentiment that Agile weakens the ability to hold contractors (and their clients) to account for the delivery of fit-for-purpose products. Whether as a result of mismatched expectation, fuzzy requirements, incompetence or downright fraud.

Our defence against these–particularly the latter two–has traditionally been the much-lambasted public procurement process. As Anthony Zacharzewski put it with characteristic tact and style, we should have a care about throwing out the defences that these processes are designed to provide, for all the tales of woe that can also be laid at their door.

Stepping back a little from all this, I am left wondering what points are really being argued here? Is this about a methodology, or is this about ensuring that the right people are getting through the door? Amazing things can happen when the usual barriers are thrown down and the real, trusted experts are brought in. (I hope to see real-life evidence of some of this next week.)

After all, with amazingly insightful and competent people involved, respecting each other, listening to reason, taking risks where necessary, being flexible–and above any suggestion of conflicted interest or perverse incentive–even the most traditional requirement/specification/build approach is going to perform pretty well. And just imagine the horror of a perversely-motivated behemoth muscling in (let’s call them Anders*nAgile, say) with their Chicago-schooled ScrumBizAnalysts(TM) charging a couple of grand a day to dance in a slightly different style to the same old tunes.

If this is about trust–and I think a lot of it might be–then we need to be careful not to confuse “method” arguments with those that are more about “gatekeeping”. A great sage in the public sector IT world once said that the only HR rule you ever needed was “No muppets”. He had a point.

Central or decentral?

Yes, nice easy question. Should be a short post.

One of the debates that stuck in my mind at the UK GovCamp 10 came from a session hosted by Alastair Smith. Ostensibly about the ‘UK snow’* and what that had meant for the likes of local authorities in delivering services and information. At least that’s what I think it was about. One can never quite tell with unconferences.

The difficult issue of managing information in disrupted conditions. One of my favourite subjects, be it weather, strikes, train disruptions or pandemics.

“How to tell people about school closures” is an excellent example.

Why’s it so difficult? Here’s a little list:

It’s a highly localised decision. It’s taken by the headteacher of a school, often at short notice. What if they’re stuck in snow, or can’t communicate their decision to anyone? We’re talking about disruption here, remember?

It’s highly time critical: if the information is to be useful it has to be delivered in the very tight window between decision and parents’ departure for school (or rearrangement of childcare, or whatever) and almost by definition this will be outside normal working hours.

There are no obligations or penalties associated with how well it’s done. (There may be a motivating issue about OFSTED reporting of absence, but I consider that secondary to the actual information process, so am discounting it from this analysis.)

There is no consistent, expected place to find the information. In some areas schools brief local authorities, in others local authorities brief local radio, there are numerous instances of online information, but little in the way of standardised approach.

Kids are involved. Kids who may just have a conflict of interest were there to be any opportunity to game the information. Just possibly.

A variety of tools are used to try and get the message out: from notifications that are actively sent to parents (by SMS, email or phone) – so-called information ‘push’; to information made available for consumption (by web, radio or pinned to the school gates) – the ‘pull’ side. Some parents and schools have developed cascade networks, formal or informal, to pass on the message. Others haven’t.

Do we have any plus sides? Well, the only one of note is that snow closure is usually predicted, to a greater or lesser extent. Something I suspect that fuels even more ire when information management fails. Surely, we cry, they must have know this might happen? Why weren’t they prepared?

Accustomed behaviours are highly personal. Parents have become used to a particular information channel, be it the radio or the web, and any changes to that will cause even more confusion, at least at first.

All complex stuff – did someone say that public service information management was easy?

But where the GovCamp discussion got most interesting was when we tackled the nub of the problem – the overarching philosophy of whether it was worth trying to centralise information at all in such circumstances. Even at the highest level, opinion is divided between attempting to centralise so that information can all be consumed in one place, and ensuring that it is maintained as locally as possible to guarantee its speed and accuracy.

For there are classic trade-offs in this decision. There is no unequivocal ‘right’ answer.

Get it to a central point of consumption (or data feed that can be consumed elsewhere) by whatever communications protocols and brute force pressures you can: advantage – easy to find; disadvantage – very difficult to make foolproof, prone to error.

Or keep it distributed, and make it easier for people to get closer to the source of the decision to get the most accurate picture: advantage – saves money, fast-when-it-works, accurate; disadvantage – hit-and-miss, accessibility, findability.

The list of challenges above should make it clear why this is far from the trivial information management problem that some might assume. One chap in the GovCamp session maintained that all it would take would be a firm hand of authority to be laid on headteachers to comply (“or else their school would be assumed to be open”). I fear that view represents a hopelessly outdated approach to getting things done that actually work.

I’ll come off the fence. I think the answer to a problem like this doesn’t lie in ever more sophisticated linking and aggregation. Building big central solutions, even with a grass-roots crowdsourcing component, probably isn’t going to work.

Instead, my experience and my gut are combining to suggest that local is the place for this information. Ubiquitously local – on school sites, via SMS, on the radio, via local authorities. Keeping them in step is the challenge: but a challenge that’s more worthy of effort than building elaborate information pipelines and monumental repositories.

*if you’re wondering why this phrasing is used, there’s some background here – which might also show why I’m so interested in it.