We often make assumptions when it comes to designing, modelling and taking decisions. Assumptions are made because sometimes we do not simply have credible data, complete information about the subject or deep enough a knowledge. Assumptions become our best guestimates. Many times, time and resources constraints (both financial and human) compel us to make assumptions and just move on with decisions. Nothing wrong.
Many analysts state the assumptions (done often in an apologic tone! But a practice so ethical and good), and some generate scenarios (optimistic, pessimistic as examples) that lead us to taking better decisions. Unfortunately, some don’t just care and leave to the reader or user of the outcomes.
Sometimes, we keep making assumptions over a long time, over decades of practice and so we don’t even realize what assumptions we are making! That is some concern.
Take example of designing of urban water distribution network. For designing this system, we first lay down network of pipes and nodes along with road network and superimpose the land use. We then allot a zone around each node for lumping the water demand that has mixed land use of today (or the most recent one). This land use around the node is often assumed to remain the same for next 20 years for which is network is to be designed.
We account for this fallacy by projecting population in the zone using a model that mimics the past growth of the city (not at node level) based on some limited data. In-migration or changes in Floor Space Index (FSI) are difficult to accommodate in these projections.
When it comes to projecting commercial and industrial demands, we run into difficulty and so we make further assumptions. Well, then we start assuming per capita water demand and follow the “norms” as applicable. This factor (e.g. 200 lpcd) is often held constant over next 20 years irrespective of changes in water use due to technology, availability, affordability and pricing of water.
We assume that the demands are concentrated at the nodes and are not exerted along the pipe (as is the actual reality). We apply a peak factor to get peak demand which is another major assumption. We further assume that all demands at the nodes are acting simultaneously! Finally, we use for hydraulic calculation (i.e. to calculate flow in pipes and pressure at node) a model such as Hazen Williams with a magic coefficient C that is itself built on several assumptions!!
And one more final assumption we make –tacitly so – that water supply equals water demand.
The net result of all these assumptions can be elegant as we feel that we solved the design problem finally – getting across all the hurdles through assumptions! We see several software tools today that spew detailed outputs, plot pressure contours and provide what if scenarios if pipe diameters are changed, valve setting is changed or reservoir height and location is shifted. We also have software codes that optimize the design for costs. I myself did one called LOOP Version 4.0 for the World Bank way back in 1990. But all this makes a sense for making publications in the Journals I suppose.
Design of urban water distribution network was one example of “cascading assumptions”. But umpteen such examples exist in environmental systems such as modelling of pollutant dispersion, assessing health risks, conducting Life Cycle Assessment, designing common industrial effluent treatment plants etc. I would urge students reading my blog to “research” on some of these examples to pull out the range of assumptions we make. It can be both revealing and inspiring to address.
It’s the learning from actual operation of the system that counts. More you learn about the behaviour of the system in practice, more data we collect and apply good analytics to the pooled information, discuss performance through roundtables and share the outcomes, more we will become knowledgble to make better assumptions. Imagine, if we processed the actual data from some of our operating water distribution networks, how confident we will be when engaged with network redesign, strengthening or improving on the operations. A practice we rarely follow today in India. This is where research and practice should merge.
There is nothing wrong in making assumptions. Assumptions are inevitable. We have to know however more about what we assume, their consequences and keep improving on our judgements by examining the reality. We need to be transparent and responsible in disclosing assumptions and learn how to live with them.
(image sourced from http://www.bing.com/images/search?q=assumptions&FORM=HDRSC2#view=detail&id=2534677EB2838CC5A081E91EAD1301BD004D9A94&selectedIndex=77)
I liked your blog about the practice of assumptions.A well written article. It’s also prevalent in other fields of practice like finance and banking. Whilst evaluating a proposal to finance a new project, several assumptions and scenarios are compared. However, these only throw up a lot of confusing numbers and the decision is ultimately taken on the basis of human emotions; that is fear or greed. We tend to side with the one which we think is most beneficial for us at that time. I think that along with assumptions, one must also mention the rationale that was behind the final decision (that would be risky, I guess).