3 ms·
Digital Government is one of the great challenges of the next decades, and I am worried that we still see software as a "rockstar" can solve it profession. Con
by lifeisstillgood 10y ago
Digital Government is one of the great challenges of the next decades, and I am worried that we still see software as a "rockstar" can solve it profession.
Conways law means that changing software to meet our needs now means hanging the whole organisation. And changing government is a whole order of challenge greater than in private sector.
But I see daily the problems of scrum sprints forcing poor architectural decisions that need strong coders to push back on - and while is suspect Matt would be one of those pushing back, really the system of digitisation of government should be more - sympathetic to the challenge.
I am amazed and impressed by the XDS approach (gov.uk is a leader in this) but I remain under convinced and unable to express this coherently - I will sleep on it
- mmahemoff 10y agoI share your concern, but how else does radical change on this level happen without a wave of rockstars (as seen with UK's GDS)? One or two isolated rockstars, or even a team of a-bit-above-average engineers, have a negligible impact when faced with a bureaucracy. Their efforts are hampered and their results can easily be dismissed as noise. Is a team of rockstars alone sufficient? No. It still needs the kind of empathetic engineers who want genuine improvement and will work to permeate their efforts through the organisation. Rockstars by definition are outstanding technically, but their mindsets can vary widely. Rockstars of the precocious, arrogant, variety aren't suited for this, but others (like Matt Cutts imo) are.
- lifeisstillgood 10y agoHow Else is a very good question - and I fear that dropping in a "wave" of anyone is just going to end up wits one wet beach - GDS is having inspirational effect but as a private contractor who is making every effort to break into (for want of better term "let me get paid for developing open source solutions to the 2000 government needs - see http://www.oss4gov.org http://www.oss4gov.org) it is clear that they are not breaking down every barrier - there is a lot to do. I am not saying I know of a better way - but rock stars don't change organisations. Meaningful change is really hard and conways law is a two way sword. If you can't change the organisation good luck having the software change it for you. Here is a longer thought : Let's look at UK schools and the department for education. The essential jobs of the department are 1) school standards and inspection (ofsted) and 2) paying the schools and so forth. Fundamentally we don't need a whole department to funnel money from treasury to headmasters. There are many pieces of software that Matt Cutts could work on there, a whole career. But we could remove the need for that department and hence that OSS software by political action. This is the point I think I am making. Software literacy is soooo vital to making sure that organisations make the right choices in their software, and software is now and will be soooo deep in all organisations (they effectively become programmable corporations) that we need software "rock stars" at the highest levels of government making the highest levels of architectural decisions But when we talk about architectural decisions at government we basically are talking ministerial level policy making. And that implies that companies are going to need to have their it architecture decided by their internal political structures - which will have to become more democratic So - yeah still not making a good point. But when we want people to come in and transform digital in government, they will quickly ask political questions. And they need political answers. Even though the leverage and driving power lies in what software can offer.
- spangry 10y ago"...software is now and will be soooo deep in all organisations (they effectively become programmable corporations) that we need software "rock stars" at the highest levels of government making the highest levels of architectural decisions" This is the key issue. Having spent years working on 'IT in government', at both the cabinet policy advisory level and at the IT coalface level, I've formed the view that when it comes to IT policy, 99% of the policy outcome is determined by implementation detail. I think this makes government IT policy quite unique, and it's something policy people (like myself) can't wrap their heads around. Perfect example: the Australian 'Standard Business Reporting' programme (http://www.sbr.gov.au/ http://www.sbr.gov.au/). It's hard to fault the policy position: open up APIs for reporting to government. But they fucked up the technical implementation so badly that they are now 5 years in, $1bn in the hole, and have around 1% voluntary take-up. The relevant senior decision makers were completely IT illiterate and went with the "can't go wrong with IBM" route. Consequently, any developer that values their sanity won't come near it with a 10 foot barge pole.
- lifeisstillgood 10y agoBut what I am saying is that just as code is the design, code is the policy The policy decisions were not "open an API" but also "use JSON" and "use fog keys for signing" and "implement the VAT submissions first" The fact those policy decisions were not made is the whole point we both seem to be making
- spangry 10y agoIndeed. Often, meaningful progress on Digital Government can be hampered by short-term political imperatives. Even in countries where the public service is supposedly 'politically neutral' (e.g. Australia). You're right to worry about this being viewed as a 'rockstar' solvable problem: the problems are systemic and often have little to do with the merits of a given technical solution. I did a stint at Australia's version of the UK GDS (www.dto.gov.au) and it was a supremely frustrating experience. Inter-agency politics, poor management and, frankly, a severe lack of technical competency at all levels were apparent. I was fortunately working in a team of (highly) technical competent people, who were had experience dealing with the the government bureaucracy. By about day two we were all bringing in our laptops and wireless internet dongles because the systems we were provisioned were utterly useless (Windows 7, no local admin, filtered internet). Our server and development infrastructure was mostly on AWS and GitHub. Every single person in my unit has since either left, or been virtually pushed out of the organisation. Our preference for building up service-based infrastructure, developing a federated and privacy-preserving digital identity framework (distinct from a technical system implementation, which my team prototyped using OpenID Connect), and direct public participation (via GitHub) were at odds with the preferences of senior management. They preferred shiny new dashboards, a single government service 'portal' and a fully centralised roles-based permissions/authentication system for all citizen & business to government transactions. I wish the solution were as simple as throwing technical talent at the problem until solved. However, this is unfortunately not the case. I admire the idealism of talented engineers who are willing to get involved to make a difference. I just hope they realise that, although absolutely necessary, rationality and technical merit are not sufficient for solving this problem. If they don't realise this, they're going to have a very unpleasant and frustrating time working in government. Probably the most important attribute they'll need is 'persistence', of the same kind needed to successfully fight a guerrilla war.