A few months ago I had a hunch I couldn't let go of. Dwellings-per-hectare (the metric in almost every Local Plan density policy) tells you almost nothing useful. The same number is achievable by suburban semis, perimeter blocks, or towers. It says nothing about built form, massing, or character. And it's gameable: hit the dph target with sub-standard micro-units and you've technically complied. So I asked a question: could a combined metric envelope — dph alongside Floor Area Ratio, building height parameters (both minimum and maximum), and site coverage limits — be constructed so that it's mathematically only satisfiable by missing-middle typologies? Perimeter blocks. Mansion blocks. Stacked townhouses. To test it I used Claude (Anthropic's AI) as a structured thinking partner, stress-testing the logic, identifying the gaps, drafting policy text. What I didn't expect was that the experiment would produce something as substantial as it did: a full draft Local Plan policy, a supporting Examination in Public Topic Paper, and a Site Capacity Statement mechanism that makes density genuinely an output of design decisions rather than an imposed target. The framework uses six accessibility tiers derived from proximity to public transport, layered with character modifiers derived from a real settlement evidence base. Two things I learned along the way: 1. Parking strategy is the single biggest density variable. On a representative T4 site, the difference between structured and surface parking (combined with a height constraint) costs you 17 homes per hectare. The policy makes that trade-off visible and auditable. 2. The best question came after the policy was finished. Esther Kurland at Urban Design Learning asked whether capping the percentage of a site accessible to motor vehicles might be a simpler, more intuitive proxy metric than FAR; a 'hero metric' that captures parking, access roads, and turning circles in one measurable figure. I don't know the answer. That's an open question for the room. I presented this at a UDL session on density last month. It's not commissioned work, Stockport MBC haven't seen it, and my honest confidence in it as a complete workable policy is about 6.5 out of 10 (I happen to think it's a bit too long, for instance). But the logic survived scrutiny, and I think the question it asks is the right one. The key on making something work in similar areas is having the evidence base (like the Stockport Character & Urban Density Study produced by Planit and similar work produced for London Boroughs). Without such background evidence, you are effectively plucking numbers out of thin air. Should liveability metrics and the density controls sit in the same policy clause, or are they better kept separate? Happy to share the documents with anyone working in this space. And very happy to be told where I've got it wrong. #planning #urbandesign #housingpolicy #localplan #missingmiddle #densitypolicy #AI
Sounds interesting - particularly the idea of being challenged on your assumed replacement metrics too. Useful to be challenged on what it is your actually trying to achieve / what problem you’re hoping to solve. A lot of policy and guidance tends to build on layers of existing thinking without really going back and reconsidering the starting points. A whole host of different metrics might help to tackle the same issue like % overlooking facade or ratio of front doors to street length, people per sqkm etc. The fact that 8 storey Eixemple in Barcelona is comparable to Manhattan in population density backs up the idea height and form won’t always tell the whole story. It’s almost as if policy wording and metrics need careful interpretation and oversight by experienced, well trained professionals! (not just ai🤞)
Interesting, and would love to see it too if you’re up for sharing it? I know from some of your previous frustrations on typical low density housing schemes coming forward, that you’re looking to direct development (especially near good public transport nodes) to an increased density.
Interesting. On a couple of points: Would be interesting to see what the results are if you add in PTAL given that’s used by quite a few LPA’s now. FAR. Check out the NYC zoning code as that heavily leans into FAR which is dependant on lot size, context, additional uses etc. Might give you some further ideas. Coverage - sq ft/acre is already a metric used by quite a few people in relation to site capacity. There is a risk that something related to FAR could only drive certain types of housing typology as it becomes a simple £/sq m calculation to work out profitability which then doesn’t deliver quality
I’d love to see it. And that’s one of the big reasons why we have updated Parking What Works Where as parking really affects how efficiently you can use land.
This sounds interesting, would be good to see. When I started reading this I was going to mention UDL and gross neighbourhood density https://www.urbandesignlearning.com/videos-details?recordId=recGcLyZSyOIXhNf3
This seems very usdful, yet to read properly.
Graeme, you’re onto something. The industry can’t continue with all the applicants measuring density as coverage, eg 16,500sqft to the acre, while the planning authority counts it as 30dwh. The clash between Imperial and Metric highlights the need for a more functional measurement that surely should highlight both efficiency and liveability. In several spatial planning systems round Europe they set a quantum of development for a site. Would love to see what you’re proposing. At this moment in time I can see lots of apartment towers being replanned as ultra-compact low-rise so there’ll even be opportunities to stress test the thinking.