on the so-called objective paradox

(General content note: A lot of my thinking has really changed since the old days of this blog. There’s some weird, mean, and polemic stuff in there.)

“the “objective paradox” — as soon as you create an objective, you ruin your ability to reach it.” [*]

[hat tip g]

That’s very concise; I like it. Effectuation theory [pdf] has some very related ideas:

  • goal ambiguity – you don’t know exactly what you want
  • environmental isotropy – everything looks the same in every direction; you don’t know what’s important
  • enaction – the moves you make change the landscape (internal and external)

Anyway, why the objective paradox? Something something goal shielding (blindness to better goals and better strategies towards the current goal), functional fixedness (looking at what’s in front of you in narrow ways). I think this is totally a thing, and it’s fundamentally in tension with legibility, the principal-agent problem, control, prediction, planning, risk, and much more. But idiosyncratically inhabiting and navigating that tension is a route to high-variance value.

A lot of my time is spent setting up my life so I can effectively work with the objective paradox, creating systems, contexts, and spaces (timers, timesheets, flexible hierarchies of unlabeled project buckets, tracking minutiae so as to minimize its impact on my life) within which I can gather possibilities and aim them (microphenomenological workspace) while remaining open to serendipity.

Some of the stuff below is annoying fluff and some of it is excellent. I’ll just lump it all together because even the fluff has a hint of something really important in it:









virtue ethics for goals and feedback systems

Within in the span of a couple weeks, more than one person has talked to me about a desire for some sort of goal feedback system or structured goal system.

The obvious thing is a list of goals and someone holding yourself accountable to them, but that doesn’t seem to work for lots of people, including myself. Part of this, I think, is that explicating some things interferes with other things that are implicit. And some goals function better implicitly, some of the time. And lots of other stuff.

One thing that I do is not labeling my projects. In my organizational software, every single entry is titled the same, an “@” sign. This makes it easier to let projects evolve and stay fresh as you and the world change. These “freeform project buckets” are fairly lightweight and easy to mix and refactor and strategize with.

But they don’t give a sense of measurable or at least systematic approach towards one’s goals, with attendant increased motivation and clarity of choice and action.

I’ve just had the idea of a sort of virtue ethics for goals. That is, “decision-point touchstones,” or a set of “compasses,” instead of a “goal-state-to-be-mechanistically-approached.” (This is likely not an original idea.)

They could be a list, they don’t need to be well-specified; they could be brief. It might be possible to arrange them in a hierarchy, or that might not be a good idea. I’ve just started playing with this.

It seems like people have an implicit sense of whether an action will take them towards or away from a goal; they implicitly calculate a long change of nonmonotonicities and arrive at some sort of sense of “towards or away.” There just needs to be some sort of trigger to do it at the right times. And that could be having a list and intending to refer to it in moments of non-duty and non-obligation. And then doing so and making choices in reference.

And so even the most high-level goals can become something that can be used and acted-with-respect-to in this very moment. (I can actually see the possibility of hierarchies, here.)

(Note that there is still plenty of opportunity for goal conflict, false dilemmas, and all sorts of stuff that might have to be resolved with other tools. But I imagine this sort of framework is synergistic.)

In having this touchstone idea, and this relatively discrete action of referencing the goal list (and then choosing and noting towards or away), perhaps this is amenable to some sort of feedback system. And perhaps this idea will evolve. I may play with parts of this if it stands up to some initial hammering.

If you want to discuss some of this other interested people, please join the Slack chat (possibly with an anonymous email). I’ll make a new channel for this.

Using Concept Specifications


  • Introduction
  • Provisional Definition of a Concept or Term
  • Two Uses for Concept Specifications
  • Structure of a Concept Specification
    • Structure of a Concept Specification
    • Alias
    • Definitions, Positive Attributes, Positive Examples
    • “Near Misses” and Distinctions
    • Exemplification Heuristics
      • Positive Examples
      • Negative Examples
      • Precise Specification
    • Relationships
    • Relationships to Prose
  • Examples of a System of Concept Specifications

Note: The formatting of example concept specifications below is a bit messed up. They shouldn’t be “bulleted,” but, yeah, that’s otherwise what they look like.


This is a collection of brief excerpts from other projects that I thought might be useful to other people. The topic is what I’m calling a “concept specification.” I use the concept specification approach when the theorizing that I’m doing is too hard for me to do any other way (which is often).

(Note: You’ll find frequent bolding below; this is what I use sometimes to denote a concept. Given that these excerpts are old or in-progress, and especially given that they’re excerpts, the bolded terms won’t always be defined. That aside, this document is intended to be fully self-contained.)

Provisional Definition of a Concept or Term

A term or a concept is something that has meaning, sense, or gist. It’s something that can be recognized as a thing distinct from other things or from an undifferentiated background. One can signify a concept using <brackets> or bolding. One can experience the conscious representation of a concept by paying attention to its meaning. For example, what does “car” (or <car>, or car) mean? (It will mean different things depending on context or agreement.) If the meaning changes, the concept changes. In some sense, the concept is the meaning. The signifier of the concept, “car,” <car>, or car can change (e.g. automobile) while still referring to the same concept, the same meaning.

Two Uses for Concept Specifications

There are at least two uses for concept specifications.

One use is to specify a concept for yourself. Another use is to specify a concept for others.

In the second case, one might strive to have a concise definition in the concept specification, as well as a complete collection of clean positive and negative examples.

In the first case, though, the concept specification can be very, very messy and incomplete, because the concept specification is for you. You’re (temporarily) finished as soon as you know what you mean, and you’re somewhat confident that the concept specification will help you remember or refresh your memory of what you mean. And, perhaps even more importantly, a concept specification can also help you clarify and stabilize what you mean. Even a single, incomplete sketch of a positive example can dramatically clarify the usage of a word or term and create a vague-but-useful category or bucket in your mind, which can become clearer and clearer over time. A concept specification is a stake-in-the-ground, a highly leveraged, refinable starting point.

Structure of a Concept Specification

One strategy of specifying a concept is by using aliases, definitions, positive and negative examples, positive and negative attributes, and relationships.

Structure of a Concept Specification

Here is how the structure of a concept specification might look:

  • alias [alias 2, alias 3…] (old alias 1, old alias 2)
  • + definition
  • + positive example
  • + positive example
  • + positive attribute
  • – negative attribute (“near miss”)
  • – negative example (distinction, potential confusion)
  • / relationship

Here is a concrete example of a “working” concept specification. (It’s not so much intended to “make sense” in this context so much as to just give an idea of what they look like and how messy they can be while still being incredibly useful):

  • combination power
  • + can swap out words with different words in same “set” and still get meaning, in fact this is indeed one part of heat and light, or perhaps rather compositionality means you (likely?) have heat and light.
  • + requires a complete set
  •  vs Add-on which is when you have compositionality but don’t have heat and light. you can add it or take it away, but there’s not fill in the blank where multiple things can go in the blank.)
  • – vs permutation power??? sort of leads back into holophrasis… almost equals language, can have permutation power

(Numerous examples of using this structure are collected at the end of this document.)


In the structure above, an alias, alias phrase, or gloss gives you signifiers to refer to the concept. They are intended to concisely signify and concisely evoke the sense of the concept. Additionally, multiple aliases might evoke slightly different senses of a broad or vague concept (which might be intended to be vague or which haven’t yet been broken into multiple, more precise concepts).

You might retain old aliases in parentheses so that you can find the concept specification from other relationships in which the old aliases haven’t yet been updated.

Definitions, Positive Attributes, Positive Examples

The plus (“+”) signs denote what the concept is (definition or rule), or has (positive attribute), as well as positive examples that can be recognized as instances of that concept.

NOTE: A concept specification does not need to contain a definition! Positive and negative examples alone can potentially specify a concept very precisely and are excellent starting points. In addition, even if a concept has a definition, the concept specification is often greatly improved with the addition of a few positive and negative examples.

“Near Misses” and Distinctions

The minus (“-“) signs denote what the concept is not. There are at least two ways to do this: negative examples (“near misses“) and distinctions. The possibilities for what a concept isn’t are very numerous, so one might initially think that it isn’t very helpful to specify negative examples, to narrow down the meaning of a concept. But, in fact, it can be very helpful to think of concepts that are very close to the target concept and how those concepts differ from the target concept. These are “near misses.” In creating a “near miss,” you might give an example of something that’s not quite the target concept as well explain why this is the case. (This indirectly specifies useful negative attributes.) Additionally, you can list concepts that could be confused with the target concept (whether they’re “close” to the target concept or not, which they often will be, though sometimes they just have similar signifiers or aliases.) In this latter case, you’re listing negative examples and once again indirectly specifying negative attributes.

Exemplification Heuristics

When exemplifying, it can be helpful to keep two general guidelines in mind. (These guidelines are from Siegfried Engelmann and Douglas Carnine’s Theory of Instruction, as part of their strategy for faultless communication.)  

Positive Examples

First, when you’re specifying positive examples, try to make those positive examples maximally different from each other (but don’t start by trying to do that; just bang them out). This is to prevent what has been termed stipulation or undergeneralization. That is, in the absence of a clear definition, and additionally if a series of positive examples are very similar to each other, a further example that’s too different will be incorrectly rejected because the sense of the concept has unintentionally become too narrow.

Negative Examples

Second, when you’re specifying negative examples, try to present negative examples that are not only as close to a positive example as possible (as mentioned above) but are once again maximally different from each other. (Or, if you are presenting many examples during direct instruction, examples might be chosen to fall into maximally different clusters.) Maximally different negative examples help to more completely outline the “shape” of the concept.

Precise Specification

Well chosen examples can lead a listener or reader to infer very precisely what you mean by a particular concept (via triangulation and bootstrapping). If more complex concepts are hierarchically built out of attributes composed of ever more highly granular concepts, which themselves are exemplified by positive and negative examples grounded in experience, communication can become practically unambiguous. This is an ambitious goal but can be achieved if the scope of conceptual transfer is limited and the effort is warranted.


Relationships show how concepts fit together and give examples of how they might be used.

The concept specification below contains an example of a relationship piece:

  • decomposable, compositionality
  • […]
  • + implies parts
  • + can imply actual combination power or just add-on; can still have perfect coverage that can be broken down into parts. There could be one or more combination power_parts, one or more add-on_parts and even holophrasis_parts. Just a single part  automatically implies that all there is is holophrasis. Two holophrasis_parts can imply rudimentary compositionality, even combination power, but again only if coming from a set. If not coming from a set, then can only have add-on_power.
  • + achievement
  • […]

(In the example above, the underscores separate aliases that occur in sequence.)

From Relationships to Prose

When relationships start appearing and growing in your collection of concept specifications, concise, precise, and even powerful prose, explanations, predictions, and theories become can much, much easier. This is an instance of the heuristic “go slow to go fast…”

Here is a brief example of prose underpinned by concept specifications:

“Using the various moves, keep reworking terms until they form mutually exclusive_sets which can undergo combination. […] This affects the underlying sense of the implicit model. […] The concepts don’t have to be native, as long as there are sufficient examples to provide for recognition while preventing miscues and stipulation. […]”

Examples of a System of Concept Specifications

Finally, below are multiple examples of (old) concept specifications in progress. These examples aren’t intended to form a complete or finished theory. Rather, they are intended to illustrate how concept specifications can be used in practice, as messy, ill-formed works-in-progress that nevertheless interrelate and can dramatically facilitate theorizing at various stages of the process.

You keep banging away at your specifications and, over time, they start precisely linking up in surprising and powerful ways…

  • moves, explication move
  • + a move can be a warned move, a complete move, a meaningless move, and a transient move
  • warned move, an incomplete move
  • +
  • (fully) resonant move
  • +
  • complete move
  • + can be X or resonant
  • meaningful move
  • + some resonance, some stability
  • meaningless move
  • + no resonance
  • + can still be valuable if it’s a deliberate nonimposition move otherwise it could be taking a stand
  • transient move
  • partially or full resonant move that isn’t imprinted
  • imprinted
  • + has a gist with a relatively slow or nonexistent or easily stabiy renewable trace,  renewability means that is perfect coverage via holophrasis or otherwise and/or definitions possibly of the faultless communication variety. faultless communication implies stability
  • explicit coverage level/amount
  • + how much or how well explication covers implicit model including/versus X vs simply missing vs implicit symbolization incompatibility
  • simply missing vs implicit symbolization incompatibility
  • + simply missing is that you *could* add it versus having to throw out at all the words in the latter condition
  • add-on versus symbolic interlock
  • + add-on is you can just add the extra component
  • + when you have compositionality but don’t have heat and light. you can add it or take it away, but there’s not fill in the blank where multiple things can go in the blank.)
  • – vs symbolic interlock is something about parsimoniousness or compression, or that ability to start generating heat and light/more than the sum of its parts, or
  • perfect coverage, symbolic interlock
  • + can be decomposible or not
  • + if it’s not then it’s holophrasis
  • + if it’s decomposable then have either combination power or add-on
  • achievement
  • + part of reality not just lying around
  • – the opposite in some sense of the myth of the given
  • myth of the given
  • + already existing categories, already existing compositionality
  • decomposable, compositionality
  • + implies parts
  • + can remove parts without breaking it. the remaining parts still have meaning
  • + can imply actual combination power or just add-on, can still have perfect coverage that can be broken down into parts. there could be one or more combination power _ parts, one or more add-on _ parts _  and even holophrasis _ parts. Just a single part sort of automatically implies that all there is is holophrasis. Two holophrasis _ parts i guess can imply rudimentary compositionality, even a certain combination power, but again only if coming from a set. if not coming from a set, then can only have add-on (??power??).
  • + achievement
  • part (two senses!), part/piece
  • (1)
  • + a part/piece can be a holophrastic part, a combination part, or an add-on part
  • parts
  • + synonomous with “piece
  • (2)
  • + drive, motive, care, caution, concern, hope, fear, dream, stake, impulse, urge, will, consideration, constraint, duty, obligation, responsibility, ??
  • should.
  • + ??? can be endogenous/intrinsic parts? or rather self parts? or rather avowed parts, or rather on-plan parts? versus ???disavowed parts, off-plan parts, imposed parts… ??? (should probably have build out…)
  • reducable, build-up, add-on
  • + can partially explicate without doing damage to the rest of the felt sense!!!
  • + can’t have perfect coverage but has possibility for partial perfect coverage
  • + can’t have combination power because don’t have a complete set
  • complete set —> mutually exclusive
  • + implies mutually exclusive, _ perfect coverage, of all possibilities
  • – note:, for now just note that mutually exclusive can be used in other places but somehow very tied to complete set…
  • possibilities
  • + ????? behavior space, _ what could happen space, ??? plan space
  • implicit sense
  • + felt sense -ish, could be structureless tag in
  • combination power
  • + can swap out words with different words in same “set” and still get meaning, in fact this is indeed one part of heat and light, or perhaps rather compositionality means you (iikely?) have heat and light.
  • + requires a complete set
  •  vs Add-on which is when you have compositionality but don’t have heat and light. you can add it or take it away, but there’s not fill in the blank where multiple things can go in the blank.)
  • – vs permutation power??? dangerous? sort of leads back into holophrasis… almost equals language, can have permutation power
  • permutation power
  • + sort of leads back into holophrasis… almost equals language
  • – can have permutation power without
  • heat and light, more than the sum of its parts, explanation capacity, turning gears
  • + heat and light = composition power + explanation capacity
  • – symbolic interlock is very similar or overlapping, but not quite.
  • + (see production interference and so forth)
  • explanation capacity/power, prediction power, intervention power…, intervention nodes, affordances…
  • holophrasis
  • + from where you dream book
  • + points at something, sort can’t leave a single word out, somehow perfect coverage or it at least points unambiguously at something??
  • + “something something one long name for precisely this” (a quote from from where you dream book.
  • + holophrastic indeterminism – many ways to translate the same thing, see coordinate transformation
  • – can have perfect coverage with or without compositionality, but definitely can’t be add-on because nothing can be left out.
  • gendlin definitions
  • + (possibly important like direct reference, circumlocution, etc.)
  • nonincorporation
  • + a possible operation where you make a pact with system one to only play with it in system two, an artifact, a toy-like feeling. a not real. thinking with symbols is taking a stand will be considered as dangerous to system one, so system one will reject the whole activity [autoincorporation rejection].
  • [can do nonincorporation or nonimposition with others, too, “everything i’m about to tell you is wrong—screen it off…
  • autoincorporation rejection
  • + artifact in its partial state doesn’t contain everything important to system one, so the thinking with symbols is taking a stand will be considered as dangerous to system one, so system one will reject the whole activity
  • incorporation mechanism
  • + the agile catching during e.g. tacit updating, a thing judgment/concern that’s initially outside the scope of what you’re working on but is modulating it so you catch it and acknowledge it as in scope
  • + like have work project X and personal issue Y has insights or even implications for X, and having emotions or issues around X and having the technical skill to notice that that’s due to Y and explicitly bring Y into the field of consideration so now X and Y are both things.
  • + “but what abouts” (and resistance for whatever reason to bringing them into the active felt sense versus the normal bouncing to them and back)
  • + see my tacit updating post…
  • factor X
  • + taking words too literally[1,2], _ production intererence, _ too soon, _ clunkyness _ active and shifting and slippery
  • take words too literally[1]
  • + when focus too much on literal meaning of the words and it kills the actual idiolect meaning, the not quite english that interlocks with rest of felt sense or what person was saying [could be speaking and immediately or in writing and coming back to it later; gist is gone or something]
  • take words too literally[2], production interference, too soon
  • + focuses too much on the literal words so that person can’t keep speaking out of their felt sense. this is sort of production interference, where stuff was written down too soon, before a good coordinate transformation was found, a good framework that could actually explicate comprehensively without leaving the critical parts tacit.
  • coordinate transformation
  • + finding a way to say things, and in fact using wildly different words, seemingly entirely different words (e.g. poetry can describe same subtle phenom with entirely different words but still be about the same thing), where are able to actually explicate the whole thing, or good enough.
  • + i think writers know about this “writing too soon” phenomenon
  • implicit vastness hidden by a tiny logical inconsistency
  • + so like anomalies in a theory, where you have this little itty bitty thing that doesn’t fit, but in fact you might need huge logical buildout to account for it (though there might be a new paradigm that isn’t that much bigger or is even smaller that does account for it) – kind of like eliezer’s thing that the more more you explain the simpler the equations in some sense, like a universe needs only a few forces. but more to the point, just the tacit, subtle felt stuff (usually) that just doesn’t fit.
  • mutual resonance, deep mutual resonance
  • + share each other’s mental models, worldspace, and know each other do, both have the same answer to “this is what’s happening here”, actual same objects a la wilber gigagloss,