In conversation with some people working on Lean, I've been thinking about a few rules for operating in a Lean fashion. They're entirely my own, but I think they have "legs."

Some General Rules

Optimum versus minimum

Optimums are not minimums. We should always be clear what an optimum is, and that there are consequences to making something that's good to have into something that we must have. We should de-couple optimization from minimum standards. Too often, we take the optimum and make it a minimum standard, which makes it harder to do anything at all. For instance, LEED might be a great aspiration, but when it's made a minimum standard for new development, it has a very un-Lean effect. The minimum standard and the optimal standard even speak to two different estates. 

1. The minimum standard should be avoid problems that would cause general harm. General harm is a government responsibility. Minimum standards should be as low as possible while still limiting general harm. Of course, there is a threshold as to how much harm should be tolerated. If residential fire sprinklers are required in a location in which the available well-water is insufficient to provide service, then that has the effect of eliminating either new housing or encouraging cheating.
2. Harm to neighbors should be adjudicated privately, although if the adjudication itself becomes too complex (lawsuits or acrimony all over the place) then that is a general harm, which should kick it back up to the government level. For instance, if lowering a standard on back extensions to houses makes everyone unhappy, then the standard is too low, and should be revised.
3. The optimum standard should be based on an idea of what's best, which is a cultural-estate issue. Note that different groups may have different ideas about what is best, and might have competing rating schemes. That's a good thing. The best model is probably Good, Better, and Best. "Best" should not be the enemy of "better," and "better" should not be the enemy of "good."
   
   The Good can be enforced by guidelines. "Better" and "best" can garner prizes. For example, take a storefront rehabilitation project. The storefront should meet code: that's the absolute minimum, and it should be enforced by law. The storefront might then have to be in the "good" category in order to meet the local requirements, but these should be guidelines with flexibility. Perhaps there's a reason for an unusually large sign, say. 
   
   "Better" performance could earn you prizes or incentives. This is a clear step up from "good." Maybe your storefront gets an award to display. "Better" performance should in no way be required, but a culture will usually develop around it, so that it becomes the norm: the middle of the bell curve. 
   
   "Best," then, is in the top part of the bell curve. It should be recognized as such publicly. However, the "best" can also be too good for general use. For instance, a theater marquee could be in the "best" category of storefronts, but if every storefront had a marquee the result would be chaos. Therefore, the "best" is generally not to be imitated willy-nilly.

Arrangements, Platforms, or "Patches"

Arrangements require standards and human judgment. Arrangements, Platforms, or "Patches" mean the same thing: a general arrangement by which something is enabled. They must support three different things.  

1. A robust technical standard. For instance, a congestion pricing license-plate scanner shouldn't go berserk when someone has a foreign license plate. It should "fail gracefully" as they say -- e.g. a first-time warning to get a compliant sticker. Or a new building could "opt out" of the ordinary building code if  it adhered to certain very prescriptive standards.
2. A quality standard. There should be a maximum level of errors, and there should be a way to adjudicate them that's fair to the user. In construction, for instance, there should be a technical standard for fitting individual pipes to mains, but there should also be a quality standard for the pipes so they don't fail immediately.
3. Human judgment. This includes judgment as to whether to enforce an infraction, and may include regular reviews, appeals, and fall-backs in case the system fails. Maybe the neighborhood has decided that everyone should plant flowers in front, but someone decides to plant vegetables. In that case, the humans in charge have to decide whether the original idea was to use landscaping to demonstrate care (in which case vegetables are just as good) or as a visible display of class status (in which case vegetables might be gauche).

Nesting scales

In a neighborhood, there are different, nesting scales. These range from the lot up. Hank Dittmar and I discussed these. 

1. The block: front and back. The effects you have on your neighbors in the back and on the sides are strong, but limited-range. Outside of overshadowing them or annoying them, you don't affect them much. Your effects on the block-front is farther-reaching, in that it affects more properties up and down the conduit of the street in front. This might be managed by a block association.
2.  The sub-neighborhood.  If a "neighborhood" is a 1/4th-mile radius area, then the sub-neighborhood is much smaller: typically a quarter or "quadrant" of that. This is the area known intimately. This will almost always be a primarily residential area, and may exclude commercial activity. Then again, it may include a corner store.
3. The commercial area or main street.  Not every neighborhood will have a commercial area, but where one does, it should be managed separately from the residential block-level management. This is typically a "Business Improvement District," or "Special Improvement District." Note that there may be some overlap. For instance, an apartment building may be in a Special Improvement District, and its residents might be in a block club. 
4. The Neighborhood. The interests of residents and commerce diverge, so that they need to be reconciled by a higher level of governance. It may be made up of representatives of BIDs and block clubs. 
5. The City.  Legally speaking, the city must delegate authority to the lower levels. It might decide, for instance, that shopfront design should be at the BID level, or (better) it may decide that it's a city matter unless there's a particular arrangement in place: a BID. If there's no BID, then the authority stays with the city.

#5 above is a difference between "platform-thinking" and "subsidiarity." Under subsidiarity, a city might decide that a particular level of authority is always responsible for a certain decision. With "platform-thinking" (as I call it) the authority is devolved only where there is a particular arrangement, platform, or patch to receive it.

Computing and Automation

Communications and decision-making

Local communications. Since Lean hopes to manage decision-making at the lowest competent level and at the latest time, communications and decisions should be local. That's not to say that technology shouldn't be involved, but it should have a low barrier to entry: both monetary and cognitive. Many of our best citizens are not technically adept. 

1. Face-to-face communication should be the primary mode of decision-making. This can privilege representative government over direct democracy. Information can be disseminated and debate can take place rapidly, but for actual decisions, someone has to take responsibility, and anonymity is the enemy. Services such as Nextdoor.com are useful, but when a decision is fraught, rapid face-to-face debate and decisions are best.
2. Technology should support communication and decision-making, but should not abstract it from the physical world. It's certainly a good idea to use things like geotagging apps, augmented reality, and Government 2.0 (pdf) to support real-world decisions and public works. However, we should not let these tools float free into abstraction, because it can lead to trolling.
3. Any technology should be easy, inexpensive, and standard for everyone. If old Mr. Jones says he's waiting for his grandson to help him, that's a problem.
4. Finally, real-time in-person committees need to follow good procedures, so that they don't become dominated by busybodies or outside interests.

Automation

Automation should support human judgment. it must not supplant it. Certainly, much can be done by automation or rote, but nothing that requires human judgment can be.  Essentially, the point of automation should be to process large amounts of information, present it to a human being to make a judgment, and then sometimes to carry it out. For instance, a neighborhood might sample recycling bins to see what the compliance level is. That should be a "rote" process, not singling anyone out, and making a statistically relevant sample. Then actual human beings have to look at the results and decide what to do about them. Finally, solutions can be rolled out using a mass-dissemination of information. Of course the actual collection would be (relatively speaking) rote work too. Thus:

1. Once a determination is made, an automatic processes should foster human decision-making. For instance, in a congestion-pricing scheme, there should be clear notice that a fee is going to be charged, there should be a clear indication of what the fee is, and there should be the opportunity to take another course, such as parking at the edge of the congestion zone. It would be best for the information to be presented before the would-be driver makes the choice of transit, bike, or car. 
2. An automatic process should never suddenly emerge and bite the user. Note, this applies to both computerized automation and to assembly-line-like processes such as, say, a bureaucratic cycle that says you can't get a birth certificate without a social security card and that you can't get a social security card without a birth certificate. In a congestion scheme, your price shouldn't suddenly double without notice when you're inside the congestion zone. Kafka understood this.