Fil Salustri's Design Site

Site Tools


Damian's Notes on NSF

This page is an old Xiki page of Damian Rogers, in which he kept notes about Alexander's NSF. All I've done is reformat the document to render reasonably in dokuwiki.

This is a page where we take a look at an example in Alexander's NSF book about the design of a tea kettle. While we find that it is an excellent example, we think that the way Alexander has presented the requirements can be done in a better way, so as to better suit the design process. Here, we look at this and suggest ways to improve.

TODO These are raw notes that need to be studied and absorbed into other projects and pages.

Alexander's Original Tea Kettle Requirements List

Pre-blurb about the kettle

“An object like a kettle has to fit the context of its use, and the technical context of its production cycle.”

“Let us consider an ensemble consisting of the kettle and everything about the world outside the kettle which is relevant to the use and manufacture of household utensils. Here again there seems to be a clear boundary between the teakettle and the rest of the ensemble, if we want one, because the kettle itself is a clearly defined kind of object. But I can easily make changes in the boundary. If I say that the kettle is the wrong way to heat domestic drinking water anyway, I can quickly be involved in the redesign of the entire house, and thereby push the context back to those things outside the house which influence the house's form. Alternatively I may claim that it is not the kettle which needs to be redesigned, but the method of heating kettles. In this case the kettle becomes part of the context, while the stove perhaps is firm.”

  • First two sentences are clearly connected to systems; also the notion of boundary.
  • First two sentences also suggest the importance of design for lifecycle (though he doesn't include disposal).
  • Using the kettle to heat domestic water shows he's thinking about /function/ (role) more than behaviour.
  • What is missing in his examples of moving the boundaries is the justification. Moving boundaries must be done only with justification.
  • Is the last word of the 2nd paragraph really “firm”?

Little blurb about form, context, and fitness

“The form is a part of the world over which we have control, and which we decide to shape while leaving the rest of the world as it is. The context is that part of the world which puts demands on this form; anything in the world that makes demands on the form is context. Fitness is a relation of mutual acceptability between these two. In a problem of design we want to satisfy the mutual demands which the two make on one another. We want to put the context and the form into effortless contact or frictionless coexistence.”

FAS: Even I have to remind myself that his context is not behavioural or functional, but structural. It's every other form except the one we have to design.
DR: Yes, which is one of the problems here, i think.

FAS: Perhaps we need to carefully and always call it the structural context or something like that? This doesn't exclude more conceptual/abstract contexts, it just clearly distinguishes them.

DR: yes, I like that. I found myself having to also make that distinction on my thesis work when I talk about form: I have to remind people it's functional form rather than physical form.

List of requirements on the kettle

Format used below:

Top level (0): requirement as given in NSF.

  • Level 1: potential problems we see with the requirements as they are stated by Alexander.
    • Level 2: the 2 components (one each from the form and context) which are in misfit for that requirement.
      • Level 3: the balance variables (bv's) we find for that misfit))

Alexander's list of kettle requirements are on p60 of NSF.

(1) It must not be too small

  • This is too ambiguous to be a useful requirement. What is too small? Why musn't it be this size? There are certainly single cup kettles for single persons who only make one cup of tea at a time, so why then does this kettle have a requirement on size? Further, a smaller kettle actually helps to fulfill some of the other requirements on the list, such as: easy to store, easy to pick up, easy to pour, economical to heat small quantities of water, etc.
    • base - heating element (assuming of course that size is related to the idea that a small kettle is inefficiently used on a large stove top element)
      • bv - geometry
FAS: possible balance variables include: capacity, mass, physical dimensions (will it fit on the counter?)…. These are off the top of my head. The idea is to come up with an exhaustive list.
DR: I would say that we would not want to have physical size or dimensions in this requirement, else it becomes exactly the same as #4 about storing in the kitchen. I'm still not sure exactly what this requirement means or is used for and as such, have not included these components in the lists further down.

FAS: Well, they could be constraints from the environment or problem definition. EG: a 12-cup kettle versus a 4-cup kettle, based on market data of what's needed. Has to fit vertically on a standard counter, under the upper cabinets. Proportions must be such that it isn't likely to tip. These don't necessarily set hard quantitative limits on size/dimensions, but they provide some guidance. And there /might/ be regulations on this somewhere - not that we need to deal with them, but we ought to recognize that they might exist.

(2) It must not be hard to pick up when it is hot

  • This, #3, and #19 are all pretty much the same requirement. There really isn't a need to separate the one overwhelming requirement into these 3.
    • handle - hand, which may include size as well as heat transfer to the hand to prevent burning
      • bv - heat transfer
FAS: I think thermal isolation is the key here - from the phrase “when it is hot” - implying that we're not (yet) saying anything about it being hard to pick up when it's cold. That might be another (possibly missing) requirement.
DR: I was thinking that whether the handle is hot and will or will not burn you is not a function of “hard to pick up”, it is more “uncomfortable” or “dangerous” to pick up when hot. Maybe another area where we need to distinguish and create a new requirement that is better worded.

FAS: Yup; that works for me. I think it still keys off thermal insulation as a property of the interface - it just has implications to safety, usability, and functionality (letting the water get cold). Thermal insulation is the property that, because of its connection to various interface elements, causes the coupling between requirements. See what I mean?

DR: agreed. the “hard to pick up” is more focused at useability as you suggest; whereas, the “when hot” is more safety. This is an example where I think we need to split this requirement into it's useful bits or lump the functional part with the other grasping requirements and make this one only about the safety when hot.

(3) It must not be easy to let go of by mistake.

  • coupled with #2 above.

(4) It must not be hard to store in the kitchen

  • I might go so far to say that this is also far too ambiguous. Must not be hard to store in the kitchen is a relative sense on the person's kitchen that it is being stored. What about those that leave the kettle on the counter each day? Or those that have gigantic drawers for small appliances and the like?
    • outer structure - kitchen cabinet.
      • Geometry (as the balance variable).

(5) It must not be hard to get the water out of

  • seems to be a reasonable requirement
    • handle/spout - water
FAS: assuming this relates to pouring, balance variables might include: the handle (“handling system”), the geometry (long skinny kettles would have a moment problem when you tip it over to pour - assuming you have to tip it at all), the mass (other aspect of moments), size of the spout.
DR: I think you just did what Alexander has been doing: you assume it has to pour!

FAS: Hey, I started it by writing “/assuming/ this relates to pouring.” This does get us to the idea of /co-evolution/; that is, you can't do /all/ the requirements for a product (down to whatever level of detail) without also doing design. So, there are /some/ things one can do to keep it easy to get the water out without making the assumption of a spout - but sooner or later you'll hit a wall and will not be able to go further without deciding whether it has a spout or not, and that's a design task.

DR: so what's our consensus on the pouring bit then? shall we leave it as a part of design that is already done?

FAS: I propose: leave it as is for the /analysis/ part, and be clear that there's assumptions here. Then, when “rewriting” the case, we start at a higher level, and drill down, and make the same assumption he did. That way, we show how the co-evolution works, without having to actually design the kettle.

(6) It must pour cleanly

  • seems to be a reasonable requirement, but also coupled with #5 above. Further, this assumes it has to pour, which is generally a no-no
    • spout - water
      • bv - geometry
DR: what i love about this one is it is basically taking the piss on all current tea pots that you can't pour a cup of tea from without spilling half of it across the counter!

(7) It must not let the water in it cool too quickly

  • I'm not sure why this is stated. The purpose of a kettle is to boil water for making a pot of tea, or a cup of tea (or perhaps something that isn't tea at all). The point here is that it is a heating vessel, not a hot water storage vessel. The hot water is boiled, transferred and then stored in a different vessel (such as a tea pot).
FAS: Unless it's a combined kettle & teapot.
DR: but if that were the case then it should be defined somewhere as such, but Alexander has specifically said that he is designing an instrument to “heat domestic drinking water”, not to “brew tea”

FAS: good point. Make sure we're clear on this in any writing about the kettle. This will definitely make a good paper.
FAS: Might relate to noticing that the kettle is ready.

DR: Right, so perhaps it implies a coupling with the 'warning system'

FAS: Say it has an automatic-off feature. You might not use the water immediately cuz you're in the middle of doing something else.

DR: I see your point, though one could argue that due to its high heat capacity, water itself will not cool quickly when left to its own devices, so then why do we need the requirement? Another grey area where I'm not sure where the requirement came from

FAS: Water is rarely left to its own devices. Anyways, there's also a disconnect between the kettle being “ready” and the kettle shutting itself off. That is, does the kettle shut off to make itself ready, or does it shut itself off because it is ready? …And will the kettle want to be paid overtime? (Sorry, couldn't resist.)

DR: valid points, though I tend to feel these are leaning more towards the same idea as req't 20; “can't boil dry without warning” and not so much on this heat loss bit. Keeping on the “cooling too quickly trend”, I'd say that hot water left in a heating device will stay hot/warm for quite some time and makes this a poor req't.

FAS: Yes, but it still forces at least the issue of some kind of thermal insulation not for safety but for basic functionality. I think thermal insulation is the BV that couples a whole bunch of misfits/requirements together. Indeed, this convinces me more and more that a way out of obnoxiously complex networks of misfits is to decompose it into the BVs.

DR: I think my point is more: is thermal insulation needed in a kettle? Most common kettles do not have thermal insulation and my point above about letting hot water sit in a closed environment is such that the water will take quite some time to cool even without the insulation. So I guess my argument is that thermal insulation is not needed in a kettle and therefore I deemed this requirement unnecessary.
DR: I agree that complex networks may indeed be simplified if we looked at the BVs. I wonder if there is some demonstration we can use for this theory?

(8) The material it is made of must not cost too much

  • seems to be a reasonable requirement
  • should bunch together with 11, 12, and 13 to form “manufacturing cost”
    • structural materials - market
      • bv is cost
FAS: Cost is the balance variable here - what it sells for versus what the user can afford.
DR: would it be better to say manufacturing cost vs. what the user can afford? since 'what it sells for' can have no relation to its actual cost of production

FAS: Mfg cost is only part of the user's cost. Not sure. I mean, yes, material cost is a manufacturing cost, which only vaguely relates to user's cost. Might I suggest as a better requirement: the material must not cost so much as to adversely affect the user's cost. That is, using a slightly more expensive material may yield a kettle that costs less overall (local v global optimum thing).

DR: I think we've both agreed that this requirement can be bunched together with 11, 12, and 13 as well and call them all “manufacturing cost”
FAS: Yes; I see this as as example of Alexander suddenly dropping down to a level of detail that is inappropriate without having made certain higher level requirements and decisions first. For example, one might set the overall cost of the kettle based on market research, and then let the designers use, like, a cap-and-trade system or something to share the cost among them.

(9) It must be able to with stand the temperature of boiling water

  • seems to be a reasonable requirement
    • structural materials/components - heated water
      • bv - temperature/heat transfer
FAS: possible balance variable: operating range of kettle v. temperature of heated water. Notice the water is part of the context.

(10) It must not be too hard to clean on the outside

  • seems to be a reasonable requirement
FAS: possible balance variables: geometry, material (easy to clean without wear).

(11) It must not be a shape which is too hard to machine

  • assumes that the kettle will be machined, which may or may not be the case. Especially considering that most kettles currently on the market are in fact made of a plastic, which is not machined, but rather moulded. Rephrasing this requirement to: /it must be a shape which is easily manufactured/, would better fit.
    • geometry of parts - manufacturing process
      • bv - manufacture
FAS: You're right. We can change this to “manufacture” which is more inclusive in our version.

(12) It must not be a shape which is unsuitable for whatever reasonably priced material it is made of

  • seems to be a reasonable requirement
    • geometry of parts - manufacturing process
      • bv - manufacture
FAS: possible balance variables: material, manufacturing.

(13) It must not be too hard to assemble, since this costs man-hours of labour

  • seems to be a reasonable requirement. Though, we'll combine with others for 'manufacturing cost'
    • manufacturing cost - manufacturing process
      • bv - manufacture

(14) It must not corrode in steamy kitchens

  • Not sure what he is on about with steamy kitchens in #14. Generally the kitchen itself is not too steamy, but rather it is the kettle which creates the steam. Perhaps rephrasing the requirement as, must not corrode in normal operating conditions, would be better
FAS: In our version, we can generalize this to something like “materials must withstand detrimental environmental forces like corrosion” or something like that. This would motivate gaining a better understanding of the operational environment before proceeding with the design details.
DR: agreed.

(15) Its inside must not be too difficult to keep free of scale

  • seems to be a reasonable requirement
FAS: Possible balance variables: material, geometry, maintenance procedures.

(16) It must not be hard to fill with water

  • seems to be a reasonable requirement
    • access size/shape - water
      • bv - geometry
FAS: possible balance variables: size of access (might be spout, but might be another opening too).

(17) It must not be uneconomical to heat small quantities of water in, when it is not full

  • As stated above, #17 seemingly contradicts #1 with the limitation on small size.
FAS: Not really, as one will tend to buy a kettle that can boil a nearly maximal amount of water. Our coffee maker makes 12 cups at once, but we rarely use it for that. We bought it so that we'd be covered when guests come over.
DR: i see your point, though the limitation on small size from #1 precludes the idea of a single cup coffee brewer for instance, but a single cup would satisfy this one.
FAS: possible balance variables: operating cost (cost of heat power) v. budget of owner.

FAS: And what is “too small”? Depends on what the user's want, which is presumably what market researchers found out and explained to the designers.

DR: exactly my point in my comments about the 1st requirement! Here also we can ask what is a “small quantity of water”?

FAS: I think this could be the lower bound of the interval defining a balance variable from the form side: minimum and maximum reasonable amounts of water. Presumably it could be different than the amount of water possible from the context side. Less than the minimum would be a problem because one could damage the kettle running it (effectively) dry.

(18) It must not appeal to such a minority that it cannot be manufactured in an appropriate way because of its small demand

  • It seems like this is trying to say: “it must not be an expensive one-off”, but the wording is rather confusing. Perhaps, /it must be marketable/ is a better choice of words.
    • manufacturing cost - market
      • bv - marketability/cost
FAS: Yah, I think that's what he meant. In the 1960's nobody gave a sh!t about “market”. :-)
FAS: can combine this with others or on its own to be “must be marketable”

(19) It must not be tricky to hold that accidents occur when children or invalids try to use it

  • is coupled with 2 and 3 above. Though I must ask why invalids and children are using something potentially dangerous such as a kettle of boiling water
FAS: I think this is just an attempt to do “inclusive design.” We have more politically correct language for it these days. Connects to marketing too.
DR: yes, i see that. Though, you could argue that “invalids” is far too inclusive a term. An “invalid” may be someone with no hands or even arms, then how do you satisfy this requirement? And a child may simply be too short to see over the counter to properly grab the kettle, which again is no fault of the kettle design.

FAS: Right. But again, it was the sixties. Watch a few episodes of Mad Men - that'll give you a sense of how f*cked up things were when I was a kid.

(20) It must not be able to boil dry and burn out without warning

  • Assuming that this means that it has some sensor system and/or warning system. In older kettles this can be likened to the 'whistle' that blows when the kettle boils, or current electric kettles which shut off when the water boils. The problem with the whistle-type is that if you forget to place the whistle on, then it does nothing and can still boil dry without warning.
    • warning system - boiling water
      • bv - ?
FAS: Remember: 1960s.

(21) It must not be unstable on the stove while it is boiling

  • assumes that it is a stove-top kettle. This assumption already assumes a design decision has been made, therefore rewording the requirement to /It must not be unstable while in use/ would be better.
    • base - stove element
      • bv - geometry/manufacture
FAS: possible balance variable: geometry, mass.
DR: I added manufacture since it is quite conveivably a manufacturing deal whether or not the base is flat and stable.
FAS: For the sake of the example, I think it makes sense to separate out all assumptions we've found. Then we say that we can't know exactly what Alexander was thinking. We can give him the benefit of the doubt and say his design problem was such that the assumptions were well-founded. I also think we should stick to his assumptions as much as possible, to keep our version consistent with his. Maintaining consistency will be very important. We can, of course, indicate that relaxing those assumptions would open up many other design possibilities.

List of Form/Context Components

FAS: I would urge you to rewrite the components in functional form rather than structural form. Might sound tedious, but I think it's important to keep the reader acutely aware that we're dealing with the design at the functional level and that doing that is a good thing. “Geometry” I think is okay, but things like “handle” need to be changed to “handling system” or some such thing. Also, it's fine to leave this list as the list of things that emerge from Alexander's text. But then we need to say these “forms” are bad because of all the biases they embed in the problem. So we then rewrite the components in functional form as part of our “rewrite” of the problem.
FAS: Also, a task to be done is to map (diagrammatically, maybe with a cmap) the connections. We should start with a misfit network as Alexander might have done it. Then revise it in terms of getting rid of forms for functions, connecting up the BVs as the agents of couplings, etc. This may require a few different versions as we may discover some visualizations just don't make sense only when we actually try them out.
DR: The diagrams have been made as you know, would you like me to add them to this page so we have them here for easy reference?

Form Components

Numbers in brackets signify the requirements that the specific component relates to.

  • Base (#1, #21)
  • Handle (#2 ,#5)
  • Spout (#5, #6)
  • Outer structure (#1, #4)
  • Structural materials (#8, #9)
  • Components (#9)
  • Geometry of parts (#11, #12, #16)
  • Manufacturing cost (# 13, #18)
  • Warning system (#20)

Context Components

  • Hand (#2)
  • Kitchen cabinets/counter (#1, #4)
  • Market (#8, #18)
  • Stove element (#1, #21)
  • Manufacturing process (#11, #12, #13)
  • Water/Boiling water (#5, #6, #9, #16, #20)
    • There is some discernable difference between water from a tap (cold, room temperature, or hot) and the boiling water after the heating process, which appear in different requirements. They all come down to the water, but in different scenarios. Hence, I have listed this contextual component as 'water' in general and also 'boiling water'.
    • We could go a step further by separating the actual properties of the water that are interacting with the form. IE: for the “pouring” and “filling” requirements, we might be looking at the fluid dynamic properties of the water; while for the “boiling dry” or “cooling too quickly” requirements, we would be looking at the thermodynamic properties of the water. I suggest we decide whether to package all of this up to “water” or separate it into the different pieces as above.

List of BVs

  • Geometry (#1, #4, #5, #6, #16, #21)
    • Bending moment
    • inlet flow
    • outlet flow
    • stability
    • capacity
    • feature location {RED what's this?}
  • Heat transfer (#2, #7, #9)
    • heating element to vessel
    • vessel to water and vice cersa
    • vessel to user
    • vessel to environment
    • water to environment
FAS: I'm wondering if we can be more specific. e.g. thermal conductivity at various other interfaces. If you think of a system architecture, one might represent it with various heat inputs and outputs; one from external heat source to kettle, from the kettle to the water (that one's internal, between subsystems, so maybe ignore it at the level of the kettle-as-a-whole), kettle to hand (at “handle system”, if one touches the side of the kettle, etc).
DR: We talked about this at the last meeting, are you still ok with the way it is presented or want to look at it a different way?
  • Cost (#8, #18)
    • market
    • scale
    • manufacturing cost
  • Manufacture (#11, #12, #13, #21)
    • tolerances
    • manufacturing process
    • manufacturing rate
    • imperfections

List of Alexander's Assumptions

We will list here the assumptions we find within Alexander's list of requirements (that is, what we presume were his assumptions based on how he has listed the requirements).

Alexander assumes that:

  • the kettle has a handle
  • the water must “pour” from the kettle
  • the kettle must be heated via stove top
  • invalids and children will use the kettle
  • there is no market for a single cup kettle
  • the kettle must be machined
  • there is no automation in assembly
FAS: This might be best given earlier in the “derivation” of a new/better set of requirements.

List of misfits derived from the above requirements

“I have deliberately filled a page with the list of these twenty-one detailed requirements or misfit variables so as to bring home the amorphous nature of design problems as they present themselves to the designer.”

“All these misfit variables are stated here in their positive form; that is, as needs or requirements which must be satisfied positively in a properly functioning village. They are, however, all derived from statements about potential misfits: each one represents some aspect of the village which could go wrong, and is therefore a misfit variable in the terms of chapter 2”

FAS: Ahhh, the positive form!

“If what we can do physically about one requirement in the form inevitably affects what we can do about the other (whether positively or negatively), we call the (misfit) variables linked.”

  • It is more acceptable to think of the requirements as /coupled/ rather than /linked/ as Alexander says.
  • It is shown above that Alexander lists his requirements and misfit variables and uses the two terms interchangeably. These two should in fact be different.
    • misfits are objective by nature, they occur and we see the imbalance, which is subjective to our perceptions. Requirements are what the customer/human needs or wants as a correction of the misfit.
    • Misfits define the bounds of the problem space, requirements define the bounds of the solution space, the forma and context define the bounds of the fitness space.
FAS: So maybe, they're converses? IE: Universe - misfits = requirements?
DR: I am thinking more: misfits + requirements = null or balance
  • The interaction of the universe and the current form gives us misfits
  • The misfits then define a problem space
  • In order to solve the problem/imbalance we create requirements to fix the misfits
  • Requirements then form the solution space
  • In solving the imbalance we create a new form
  • The new form is measured against the misfits, forming a fitness space
  • Minimizing the fitness space fixes the misfits
  • Less misfits means less imbalances and the problem is “solved”
  • The new solution is a form
  • The interaction of the new form and the universe shows us if there are misfits → go back to premise 1.
FAS: You're right.

Problems, etc we see with that list

FAS: Next step: look for coupled requirements/misfits. For each set of coupled misfits, figure out what design variable(s) (e.g. the handle, the quantity of hot water, etc) cause the coupling. Do this only of a few coupled misfits, then let me know.

Revised List for Better Design


research/damian_s_notes_on_nsf.txt · Last modified: 2020.03.12 13:30 (external edit)