Infrastructure transparency · a portal on life support

The portal is online. Diagnose why it is dying.

A disclosure portal is not a website. It is a public window on an internal workflow, kept alive by five connected organs. Look inside, connect each one, and watch whether the body lives or dies over three years.

Brain Blood missing Spine missing Immune missing Muscle missing Metabolism missing Online. Looks healthy. The vitals say otherwise. Look inside →
The face, and the vitals beneath
Data freshness Last updated 498 days ago
Delivery evidence Missing (contract value only)
Record vs reality Unchecked
Public usefulness Low (nothing to do here)
Operational funding Grant closed, no owner

The homepage looks healthy. The pulse does not. The body is only skin-deep from here, switch to X-ray the organs to look inside.

The patient
This portal is technically online. Its contract values are visible. But nothing reaches the public as usable accountability. Click an organ to see why, then repair it.

Infrastructure transparency · portal on life support

The portal is online. That is not the same as alive.

Five organs keep a disclosure portal alive. Set each one and watch its three-year prognosis change.

Patient vitals
Data freshnessLast updated 498 days ago
Delivery evidenceMissing (contract value only)
Record vs realityUnchecked
Public usefulnessLow (nothing to do here)
Operational fundingGrant closed, no owner
Repair the organs
Bloodstream
Data extraction
No data moves. The portal shows whatever was typed in once, long ago.
Spine
Procurement to delivery
The record stops at contract award. There is no line to the building site.
Immune system
Independent review
No one checks the claims against reality. Optimistic figures face no consequence.
Muscles
Real user tasks
A carousel, PDFs and a press-release archive. No task anyone can finish.
Metabolism
Operational funding
A launch event, a consultant, branding, one-off training. Then nothing.
Prognosis over three years
With none of the five connected, the portal goes dark by year three. Connect the organs and the prognosis changes.

Portals do not die on the homepage. They die inside the workflow.

Every portal that launched to a press release went dark within three years. The ones that lived shared one trait: the data was a byproduct of the work, not an extra task.

The five decisions are not features to add later. They are the architecture, chosen at the start. Automated extraction keeps the data fresh without anyone logging in. OC4IDS links the contract to the building site, so the public sees the roof, not just the receipt. Independent review tests the record against reality. A practitioner task gives someone a reason to return on a Tuesday. And funding the pipeline as architecture, not a project, is what carries it past the grant.

Skip any one and the portal can still launch beautifully. It simply will not survive contact with the workflow it was meant to open.

Methodology and data

The Portal ICU is an explanatory model, not a calibrated forecast. The five survival decisions (automated extraction, OC4IDS procurement-to-delivery linkage, CoST Independent Review, practitioner-first design, and a sustained funding pipeline) are taken verbatim from the source analysis of five disclosure-portal implementations across Uganda and East Africa. The relative-importance weights express how much each decision protects a portal over three years; they are a modeling device to drive the simulation, not measured survival probabilities, and the three-year curve illustrates causality rather than predicting a specific outcome. OC4IDS is the project-level standard that links a contract to its delivery and completion; where contract-level detail is referenced it lives in the linked OCDS release. The CoST Independent Review (assurance) is a real mechanism. The "498 days", "UGX 2.4B" and "82% complete" figures are illustrative of the failure modes described, not specific to a named project.

organdecisionrelative_importance_weightfailure_if_missingconfidence
Bloodstream (Data extraction)Automate data extraction from source systems30 of 100 (modeled, not a survival probability)The pipeline clogs in spreadsheets, email and forgotten uploads. Freshness dies.Observed across 5 implementations (Uganda and East Africa, as reported)
Spine (Procurement to delivery)Link procurement to delivery with OC4IDS22 of 100 (modeled, not a survival probability)The public gets the receipt (UGX 2.4B awarded) but not the roof. Award without delivery.Observed across 5 implementations (Uganda and East Africa, as reported)
Immune system (Independent review)Run the CoST Independent Review20 of 100 (modeled, not a survival probability)Record says 82% complete, good quality, no delays. Site shows no roof, no drainage, no crew.Observed across 5 implementations (Uganda and East Africa, as reported)
Muscles (Real user tasks)Build for the practitioner, not the press release16 of 100 (modeled, not a survival probability)Impressive at launch, useless on Tuesday. No one returns because no one had a task.Observed across 5 implementations (Uganda and East Africa, as reported)
Metabolism (Operational funding)Fund the pipeline as architecture, not a project12 of 100 (modeled, not a survival probability)Launch money is a sugar rush. When the grant closes, the body starves.Observed across 5 implementations (Uganda and East Africa, as reported)

The portal is not the product. The pipeline is the product.

Transparency cannot be an extra task bolted onto busy people. It has to become a byproduct of the work itself.