Reversible and irreversible decisions (2018)

Decision reversibility is much overlooked in resilience engineering.<p>In a perfect world we&#x27;d have a kind of quantum uncertainty, where we forked reality into two streams, and maintained two options until one or the other proved a safe passage. Quoting from Digital Vegan;<p><pre><code> &quot;Consider how the UK government let our drinking water reservoirs be sold off for property development, believing that advanced JIT (just in time) management technology, smart metering and so forth, would dispense with them. Then climate change came. Reservoirs are like power supply capacitors; they absorb as well as smooth out supply. Now in the UK we have housing estates built on flood plains. Rivers burst their banks with every downpour. Knocking down thousands of peoples&#x27; houses to regain reservoir capacity is much &#x2F;harder&#x2F; than it was to sell the reservoirs to developers.&quot; </code></pre> The transition from a reservoir to a housing estate looks like a net gain in &quot;order&quot; (entropy reduction) because it seems to create value, but considering the system as a whole (cost of losing infrastructure) it increases disorder.<p>Similarly our gushing project toward an &quot;online cashless society&quot; is playing with dangerous forces. It&#x27;s a net destruction of wealth and order. It won&#x27;t be cheap or quick to re-open shops, print and distribute cash, install ATMs and money handling facilities once the terrifying brittleness of a wholly digital economy becomes clear. Those imagining &quot;Nothing can possibly go wrong with the all Bitcoin + Amazon society&quot;, haven&#x27;t thought through the reality of what happens when tens of millions of people can&#x27;t get food even though it&#x27;s in a warehouse less than 100 miles from their house.

I guess russian leadership failed to consider this when they started their illegal imperial war to annex Ukraine. People of russia are almost irreversibly becoming the new &#x27;nazis&#x27; of whose attrocities the kids will learn at school and of whom they will learn to despise for the rest of of their lives (while the earlier ones will be just another chapter in the history books that is read with the same enthusiasm that most study geopolitics behind WW1 currently).<p>This is not to suggest that the attrocities (at this point anyway) would be comparable, but that history is moving ahead.<p>Sorry about political angle, but due to my circumstances, I am very angry of what is happening, and I think these things cannot be repeated too many times.

Interesting framework. Perhaps one of the difficulties is that decision reversibility tends to lie on a spectrum between &quot;reversible&quot; and &quot;irreversible&quot;, and sometimes it&#x27;s hard to even know where on that spectrum it lies. Thus you end up having to figure out if a decision is &quot;reversible&quot;, maybe even first needing to develop a decision framework in which to make such a determination. I think all of the truly difficult decisions I&#x27;ve made in work and life have been in this category.<p>Still, I appreciate this concept and I think it&#x27;s better to have the mental model than not.

I&#x27;ve used a similar model successfully in work situations where a decision is taking too long to make but is something that&#x27;s reversible: pick a path, and put a note in the calendar to review that decision in a couple of weeks time.<p>If it turns out to be the wrong direction you can course correct then.<p>This really helps if not everyone on a team is convinced that it&#x27;s the right decision: they can commit to it knowing that there will be a chance to change direction pretty soon if it turns out not to work.

I&#x27;ve been learning a bit about distributed databases and of course AWS is a major user of such systems. The decision to do a mass migration from one distributed system to another, as an example, doesn&#x27;t fit well with the kind of &#x27;executive boardroom&#x27; mentality that this article is talking about.<p>The fundamental issue is more just taking the possiblity of error and failure of strategy into account. Hence, having a rollback strategy in place beforehand is wise. This post discusses a few basic recovery strategies plus more complicated ones:<p><a href="https:&#x2F;&#x2F;aws.amazon.com&#x2F;blogs&#x2F;database&#x2F;rolling-back-from-a-migration-with-aws-dms&#x2F;" rel="nofollow">https:&#x2F;&#x2F;aws.amazon.com&#x2F;blogs&#x2F;database&#x2F;rolling-back-from-a-mi...</a><p>Bezos is likely well aware that the real world is always more complex than two simple choices, but that&#x27;s too much information for a sales-confidence pitch to shareholders, I imagine.

Spatial data makes a lot of the 1 dimensional thinking more and more of a 20th century relic and even less of an opinion. &quot;Those variables are unique to them&quot; is an easy way to rationlize.<p>The culture divide make so many variables moot&#x2F;invalid, since lifestyle, language and customs may not even associate with any of the variables (one-way&#x2F;two-way doors for example).<p>As an individual, its easy to simplify. As someone trying to explain something, its easy to oversimplify, missing any&#x2F;all of the details, experience and even the accumulated wisdom in basing the decision(s).<p>Reversible&#x2F;irreversible is not usually a carrot on a stick. Even with health, detrimental&#x2F;beneficial usually has more merit than any cost weighted variables (you can never buy your health back).<p>2=1+1 has worked just fine in traditional science for centuries. The (2) is usually unique though.<p>All that to say.... This article did nothing for me :)

I use this model a lot in life. I like to remember about this when in arguments with someone - once you say something that you later regret, it&#x27;s very difficult to take it back.

I try to apply a similar mental model to new purchases. It can be tempting to weigh all the pros and cons of different products and find out a detailed list of all the possible alternatives.<p>But often you just have to try a product to find out if it works for you. In that case it can help to think about what risk there is in just buying something that seems ok.<p>If you can return it easily, or if it doesn&#x27;t lose much value in using and you can sell it used for 90% of the new price then there&#x27;s no real downside to just making a quick decision.<p>Detailed analysis is only really worth it for more &#x27;irreversible&#x27; purchases where it&#x27;s either a big hassle to undo the purchase (buying a house) or there&#x27;s a large cost (cars lose value quickly).

The way I&#x27;ve framed it is the cost of making the decision should be in proportion to that decision turning out to be wrong.<p>It&#x27;s extremely rare for decisions to be as binary as one-way or two-way, but there&#x27;s always a point on the sliding scale of cost to change later.

This was a key lesson I learned creating and deploying aws services at Amazon.

This may work for business, where only money is at stake. But the hardest decisions in life are the ones where both options are irreversible. e.g. choosing to have an abortion or have a baby with someone you barely know. There&#x27;s no reversible way to check the other side of either door.