Thriving with complexity using open sociotechnical systems design
Trond Hjorteland ยท article ยท 2022-01-05 source โ Reader ๐ฅ๏ธ
Ashby's Law (Requisite Variety)Coordination CostHierarchy & Org ChartsSociotechnical ArchitectureTeam AutonomyTeam Topologies
Software organizations should undergo a radical, participatively-designed structural transition from DP1 (bureaucratic, redundancy-of-parts, externally-coordinated) to DP2 (self-managing, redundancy-of-functions, internally-coordinated) sociotechnical design โ the empirically-grounded Open Systems Theory tradition that agile, product, and 'team-first' movements have been independently and incompletely rediscovering โ because only DP2 generates the variety needed to thrive in today's turbulent technical and social environment, and because that change must be structural and self-designed, not a culture or values campaign layered on top of unchanged structure.
practical ยท mixed โ Grounds a practical prescription (ICT organizations should undergo a radical structural transition) in a theoretical tradition (Tavistock Institute / Open Systems Theory) the author treats as settled scientific research rather than opinion; expository history leading to an argumentative conclusion. Originally published at InfoQ, cross-posted to LinkedIn.
Structure
| Part | Role |
|---|---|
| Framing: the ICT industry has mastered the technical side of sociotechnical systems but neglects the social side, despite claiming people/talent as its core asset | context |
| History: the Tavistock Institute coal-mine studies that discovered semi-autonomous work groups, and the parallel to agile/XP's rebellion against waterfall | evidence |
| Theoretical foundation: Emery & Trist's Open Systems Theory, the four causal textures of organizational environments (placid random/clustered, distributed reactive, turbulent field), and today's ICT environment as Type IV turbulent | evidence/context |
| The two genotypical design principles, DP1 (bureaucratic) and DP2 (self-managing), and the argument that only DP2 increases variety to match a turbulent environment | core argument |
| Diagnosis of failed or partial transitions: culture-first change doesn't work; mixing DP1 and DP2 ('mixed mode') is a maladaptation to maladaptation; industrialized/scaled Agile exemplifies this | evidence/counter-case |
| Prescription: complete, participatively-designed structural change to a non-dominant hierarchy of functions, business-capability-owning teams, and the six psychological criteria for productive work (now codified in Norwegian law) | conclusion |
| Summary and acknowledgment of the Tavistock/OST researchers | conclusion |
Key terms
- DP1 / DP2 (genotypical design principles)
Emery's two fundamental organizational design principles, distinguished by how redundancy is achieved: DP1 gets redundancy from interchangeable, simple parts coordinated by a level above the work (classic bureaucracy, maximum division of labor); DP2 gets redundancy from multi-skilled people who can cover for each other, with coordination and control held by the group itself (semi-autonomous, self-managing teams). - causal texture / Type I-IV environments
Emery & Trist's four ideal-typical organizational environments of increasing complexity โ placid random, placid clustered, distributed reactive, and turbulent field โ used as an evolutionary staging of how turbulent today's ICT environment has become. - non-dominant hierarchy of functions
The organizational structure the author argues DP2 produces: hierarchy persists (functions still need relating to each other across levels) but without personal dominance โ management functions are distributed through self-managing groups rather than concentrated in managers who control subordinates. - the six criteria (Emery & Thorsrud)
The psychological requirements of productive, satisfying work validated during the Norwegian Industrial Democracy Program: reasonably demanding work, opportunity to learn, an area of decision-making, social support, relating work to social life, and a job leading to a desirable future โ now measured in Participative Design Workshops and codified in Norwegian law.
Propositions
- explicitSoftware development is a fundamentally sociotechnical enterprise in which the social and technical systems are not separable, so they must be jointly optimized rather than optimized independently.conf: high
- explicitOptimizing the technical side of a sociotechnical system in isolation produces a dehumanizing work environment, while optimizing the social side in isolation prevents the technology from reaching its potential.conf: high
- explicitToday's ICT industry operates in a 'turbulent field' (Type IV) environment โ the most complex of Emery & Trist's four causal textures โ driven by internet scale, machine-to-machine systems, and AI.conf: medium
- explicitTwo genotypical organizational design principles exist โ DP1 (redundancy of parts, coordination located above the work) and DP2 (redundancy of functions, coordination held by the group) โ and most real organizations are neither purely one nor the other.conf: high
- explicitDP2 structures increase organizational variety (adaptive capacity) while DP1 structures decrease it, so matching a turbulent (Type IV) environment requires a DP2 structure.conf: medium
- explicitAttempting to change an organization's culture or values first, independent of structural change, has repeatedly failed to produce real change; behavior and structural change must come first, with values and attitudes following.conf: high
- explicitA genuine DP1-to-DP2 transition must be a complete, organization-wide structural change, not one confined to the 'shop floor' level, because partial change leaves an unstable, maladaptive mix of DP1 and DP2 that is worse than either pure mode.conf: highextendsTeam Topologies: Organizing Business and Technology Teams for Fast Flow ยท p15 โ Hjorteland's DP1/DP2 mixed-mode analysis supplies the deeper theoretical mechanism behind Team Topologies' more pragmatic claim that org-chart/matrix-based structuring is counterproductive โ org-chart reliance is a textbook DP1 remnant that destabilizes an attempted DP2-style (team-first) transition.
- explicitIndustrialized, scaled Agile frameworks, applied to accommodate an existing bureaucracy, exemplify the DP1/DP2 mixed-mode failure, making things worse for both the bureaucratic and the agile-team sides.conf: medium
- explicitAgile, product-organization, Spotify-model, Holacracy, Dialogic OD, and 'teal organization' movements are all independently seeking a DP2-like structure without anchoring their approach in Open Systems Theory.conf: medium
- explicitcontestedA genuinely transformed team should own a complete business capability end-to-end, with full authority such that no one outside the team has a say in how the work is done, because everyone needed is already a team member.conf: highsupportsTeam Topologies: Organizing Business and Technology Teams for Fast Flow ยท p23 โ Hjorteland's fully-authoritative, business-capability-owning team (nobody outside has a say) independently arrives, from the Open-Systems-Theory self-managing-group lineage, at treating the team as the atomic, self-sufficient unit of ownership โ the same premise Team Topologies is flagged as implicitly assuming.contradictsTeam Topologies: Organizing Business and Technology Teams for Fast Flow ยท p16 โ Team Topologies argues architecture shouldn't be designed in isolation from teams, but still envisions topology as something leadership designs and evolves (reverse Conway maneuver, sensing and adapting); Hjorteland insists the design itself must be done participatively by the people who will implement it, treating any externally-designed restructuring โ however team-aware โ as reproducing a DP1 remnant of coordination located above the work.
- explicitA well-resourced, cross-functional 'product team' with apparent authority is often, underneath, still a temporary structure embedded in a surrounding bureaucratic (DP1) organization โ e.g. governed by PMOs or an ordering-supplier divide between IT and the business โ as Spotify's own experience showed.conf: mediumextendsSociotechnical Architecture as Enabler of Product Thinking ยท p2 โ da Silva's point that teams need the right scope/capacity (topology) to actually build the intended product is given a sharper, falsifiable theoretical edge by Hjorteland's Spotify example โ apparent capacity/authority can mask an underlying DP1 bureaucratic structure that undermines it.
- explicitProductive, satisfying work requires six psychological conditions โ adequately demanding work, opportunity to learn, an area of decision-making, social support, relating work to social life, and a desirable future โ now codified in Norwegian law (Work Environment Act ยง4.2).conf: high
- explicitA structural change should be designed by the people who will have to implement it, not imposed on them from outside.conf: high
- implicitThe DP1/DP2 theoretical apparatus, developed from mid-20th-century coal-mining and mechanical-processing industries, transfers cleanly to knowledge work in software organizations.conf: medium
Arguments
ICT organizations must undergo radical, complete DP1-to-DP2 structural change โ not incremental culture change or partial local change โ to thrive in a turbulent environment. strength: medium
Explicit premises:
- Technical and social systems must be jointly optimized; neither works in isolation.
- Culture/values-first change has repeatedly failed to produce real change.
- DP2 increases organizational variety while DP1 decreases it.
- Partial change produces an unstable DP1/DP2 mixed mode that is worse than either pure mode.
Implicit premises:
- Matching environmental turbulence requires matching or exceeding it with internal organizational variety (an application of Ashby's law of requisite variety that the author invokes without naming it).
- The only way to avoid the destabilizing mixed-mode trap is to change everything at once, rather than there being a viable staged or incremental path to full DP2.
Inferences:
- If the mixed-mode trap is real, then well-intentioned but partial transformations (a single empowered 'product team' inside an unchanged bureaucracy) may be actively counterproductive rather than simply insufficient โ a stronger and more falsifiable claim than it first appears.
Agile, product-org, and 'flat'/teal movements are reinventing DP2 incompletely because they lack Open Systems Theory's diagnostic and design apparatus. strength: medium
Explicit premises:
- Several named movements (Denning's network-of-competence, Dubakov, Spotify Model, Holacracy, Dialogic OD, teal organizations) articulate goals that closely resemble DP2.
- Spotify's own publicly acknowledged retrospective found its widely-copied model was, underneath, a bureaucratic matrix structure.
Implicit premises:
- Without OST's causal-texture diagnosis and the Participative Design Workshop's measurement of the six criteria, practitioners lack a way to detect whether they have achieved genuine DP2 or only a DP1/DP2 mixed mode dressed in DP2 language.
- A single, well-publicized case (Spotify) is treated as representative of the broader pattern across the named movements.
Inferences:
- This corpus's own modern 'sociotechnical architecture' and Team Topologies material, which also pursues team autonomy and fast flow without foregrounding Tavistock/OST, may be another instance of the same independent-rediscovery-without-anchoring pattern Hjorteland describes.
Narratives
- Tavistock Institute researchers studying English coal mines in the late 1940s found that miners, confronted with new technology that imposed division of labor and external supervision, reverted to self-organizing, relatively autonomous groups โ and that this reversion improved productivity, worker commitment, and physical/mental health.
โ Functions as the founding empirical case for the claim that jointly optimizing social and technical systems (rather than optimizing the technical system alone) produces better outcomes โ the historical root the rest of the argument builds on. - XP teams in the 1990s rebelled against the prevailing huge, multi-year waterfall projects out of frustration with their inefficacy, independently arriving at practices resembling DP2's self-managing groups, decades after and without reference to the Tavistock research.
โ Supports the proposition that agile and STSD are parallel, independently-arrived-at expressions of the same underlying DP1-to-DP2 dynamic, reinforcing the 'without anchoring it in OST' proposition. - Spotify's own public 'failed squad goals' retrospective revealed that its famous, widely-copied model was, structurally, still a bureaucratic matrix organization rather than the self-managing, DP2-style teams it was perceived to be.
โ Supports the proposition that apparent product-team empowerment often remains a DP1 structure underneath, used as the article's concrete cautionary case.
Questions addressed
- โ
How can ICT companies thrive in an increasingly turbulent (Type IV) technical and social environment?
By undergoing a complete, participatively-designed structural transition from DP1 to DP2 โ self-managing, business-capability-owning teams with a non-dominant hierarchy of functions. - โ
Why do most agile/product transformations fail to produce lasting change?
Because they attempt culture-first change, or achieve only a partial/local DP2 pocket within an otherwise-unchanged DP1 bureaucracy, producing an unstable 'mixed mode' worse than either pure design. - โญ Does a theoretical framework built from mid-20th-century coal-mining and mechanical-processing research need adaptation to apply cleanly to interdependent, fast-changing knowledge work like software development?
Not addressed โ the transfer is assumed, not argued. - โญ How should multiple DP2, business-capability-owning teams coordinate with each other at organizational scale, once each has full internal authority?
Not addressed โ the article's structural prescription focuses entirely on internal team democratization and leaves inter-team coordination (exactly the problem Team Topologies' interaction modes and platform teams address) unexamined.
Gaps
- incompleteThe article's prescription for a fully-authoritative, business-capability-owning team addresses internal team democratization in depth but says nothing about how multiple such autonomous teams coordinate with each other at organizational scale โ precisely the cross-team problem (interaction modes, platform teams, Team API, fracture planes) that occupies most of the modern Team Topologies model.
Read next: Read against extract-team-topologies' interaction-modes and fracture-plane machinery (#p10, #p12) to see whether DP2's radical internal self-management and Team Topologies' designed inter-team interfaces are complementary layers or in tension over who retains coordination authority. - incompleteThe piece doesn't engage with how AI coding agents inside a team change the DP1/DP2 calculus: DP2's 'redundancy of functions' depends on human members being able to cover for each other with genuine skill, social support, and psychological ownership โ properties the six criteria measure โ and it's unclear whether or how an AI agent can participate in that redundancy, or whether agents instead reintroduce a DP1-like dependency on an external (non-team, non-human) coordinating resource.
Read next: Read against extract-latent-space-yegge-vibe-coding-manifesto's shrinking-team-size claims and extract-da-silva-sociotechnical-architecture-product-thinking#p7 on 'the team' as an implicitly fixed human unit. - uninformedThe claim that industrialized/scaled Agile frameworks make things 'worse for both the DP1 style organisation and the DP2 style agile teams' is explicitly hedged ('it can be argued') and supported by no specific case, study, or citation, unlike the rest of the article's claims which are each backed by named research or documented cases (Tavistock, Emery & Trist, Spotify).
Read next: Look for documented case studies of scaled-agile (e.g. SAFe) adoption outcomes to test this specific claim rather than accepting it on the strength of the article's otherwise well-sourced argument.
Related documents
- Optimizing teams: A dual perspective on Team Topologies
- Beyond Autonomous Teams in Software Product Development
- Scaling the Practice of Architecture, Conversationally
- How Platform Engineering Using Golden Bricks Can Enable Fast and Smooth Delivery
- Architecture Modernization: Aligning software, strategy, and structure โข Nick Tune โข CodeCrafts 2024
- AI Won't Save the Kingdoms We Built
- Software Development As Collective Learning With Hazel Weakly
- Sociotechnical Architecture as Enabler of Product Thinking
- DevOps Topologies
- The engineering manager role is splitting in two