Skip to content
CTS Field Notes

CTS Field Notes

Overlooked problems, assumptions worth questioning, and things worth knowing.

Practical technology judgment for business leaders. Fresh perspectives for curious minds.

Clear, evidence-based explorations of everyday technology decisions and the larger systems behind them. Written for curious readers who want more than a checklist.

How Work Keeps Moving When Email Goes Down

In this fictional example, a customer calls to move a delivery from Tuesday to Thursday. The person who answers can take the call. Can they also reach dispatch, find out whether Thursday is open and get the new date back to the customer while the inbox is down?

On November 12, Microsoft reported that users had trouble connecting to Exchange Online mailboxes. Its updates said core Teams messaging and online meetings were largely functional during the broader incident, while some Exchange-dependent tasks, including meeting scheduling, could fail. Microsoft reported an earlier, separate Teams interruption that day. That distinction matters if you’re trying to keep work moving. Try the handoff in the existing chat and ask dispatch to confirm that the message arrived. Microsoft incident reports reproduced by Penn

Follow the request through the business

Take the delivery change as an example. The customer already has a number they can call. The person answering needs a place to hand off the request and a way to see whether dispatch picked it up. If the usual team chat works, put the order number, requested date and person responsible for the callback in one existing channel. Have dispatch reply there once it has checked the route. If the route is full, the caller can go back to the customer with a real choice instead of a promise someone else will have to unwind.

If chat is down too, use the phone route you’ve already agreed on. For a small team that might be the dispatch desk first, then the person covering it. Give the person who takes the request ownership of the callback until someone explicitly accepts it. A voicemail can leave a routine request pending. Whoever took the customer’s call still owns the callback until dispatch accepts the handoff. Confirm a new date only after dispatch checks the route. If today’s driver is already on the road, you need someone to answer and confirm what the driver should do.

Outside the company, the best channel may be different. Your staff might share a chat channel, but a supplier or customer may never see it. If your office phones run through the same internet connection as everything else, try the fallback from a phone on cellular service during a drill. A number is useful only if someone can dial it and someone else knows to answer.

Decide where the agreement goes

In ordinary work, the email thread often holds the history. During an outage, a chat message and two phone calls can leave different people with different versions of the same delivery date. If the order system works, have the person confirming the new date update it while the customer is still on the phone. If it’s unavailable, use one designated temporary log with access limited to the people handling the work. Record the order number, the confirmed delivery date, who confirmed it and whether the customer agreed. Copy the decision into the order record when it comes back and mark the temporary entry as transferred. That gives the next shift somewhere to look before it calls the customer or sends the truck.

The Department of Energy’s September 2026 communications drill asked participants to test alternate routes with their incident-response partners. A small business can borrow the test without borrowing the utility sector’s plan. Department of Energy drill guide

Run a fictional delivery change with email closed. Tell everyone it’s a drill and keep test changes out of live orders. When dispatch confirms Thursday, the person who took the call should be able to tell the customer the new date and show the next shift where it was recorded. If the date exists only in a voicemail or someone’s chat history, the next shift still has to figure out what to put on the truck.

Permalink

A Grant Allocation Is Not a Resilience Project

On October 20, the U.S. Department of Energy announced about $967 million in additional Grid Resilience State and Tribal Formula Grant funding. For a community or organization interested in a local improvement, the next question is narrower than the headline: which part of a proposed project could this program actually fund?

The money goes to states, territories and federally recognized Indian tribes, which select eligible implementing entities through their allocation processes. An announced allocation is not an award to a particular town, business or utility. Nor does a useful infrastructure proposal become eligible simply because someone describes it as resilient.

One proposal, two funding scopes

Consider a hypothetical utility proposing to bury a vulnerable stretch of overhead line. Its long-term plan covers an entire route, but the available subaward would cover only one segment. The utility wants to present that segment as the first phase of the larger effort.

DOE’s October 28 FAQ makes the boundary consequential. Section V.5 says the federally funded portion should deliver a grid benefit on its own, without depending on other separately funded projects succeeding. The recipient must distinguish that portion from the broader project.

In our example, drawing a line around the affordable segment is only the start. Suppose its proposed benefit depends on completing another segment next year. If that later work has no committed funding, the first segment’s promised result rests on a project outside the subaward. Calling the first segment a phase does not resolve that dependency.

Now consider a different scope within the same hypothetical route: a segment whose undergrounding would address a specific exposure even if the rest stayed overhead. That proposal has a different funding argument. Its benefit can be explained separately from the aspiration to finish the route. This is a comparison of proposal logic, not an engineering finding that either segment would qualify.

The distinction could change which segment the utility proposes, how it describes the work, and what it seeks from other funding sources. It could also reveal that no useful independent segment exists. In that case, the team has discovered a financing problem before presenting a partial project as a complete resilience improvement.

Equipment labels do not settle eligibility

The same discipline matters when a proposal involves backup power. FAQ V.1 and V.4 expressly allow emergency backup combustion-based generation and specified outage-response storage. That does not make every generator or battery project eligible. Confirm the entity, technology and scope against the applicable award with the recipient and Federal Project Officer.

A facility may have several sensible continuity investments in mind: power equipment, controls, communications and staffing arrangements. Those are general planning possibilities, not a list of activities this grant will pay for. The proposal needs to separate the desired facility outcome from the particular work submitted for funding. Otherwise a persuasive description of the whole improvement can conceal an unsupported assumption about one component.

The recipient’s next step also depends on its history

A new applicant and an existing recipient do not necessarily repeat the same process. Under FAQ I.5, an existing recipient needs a revised narrative and associated notice and hearing when its implementation approach changed or its prior narrative did not contemplate projects beyond the earlier funding years. A new allocation alone does not automatically require another hearing.

For the hypothetical utility, that means the conversation with the sponsoring recipient should address both the proposed segment and the allocation process that applies now. A national announcement cannot answer either question for the local proposal. Nothing in this example establishes a Michigan award, a selected project or a local operating result.

The useful decision is whether to propose a fundable segment with an independent benefit, secure the missing support for a larger undertaking, or reconsider the scope. That is a more consequential choice than treating the size of the national allocation as the size of the local opportunity.

Permalink

How the 2026 Midterms Could Change IT Planning

AP projects Democratic control of both chambers, according to Axios. For IT professionals preparing next year’s budgets, that raises immediate questions about AI rules, data-center capacity and federal cybersecurity support. The practical impact will arrive on different schedules.

A congressional majority can change which proposals advance and which agencies face scrutiny. It cannot, by itself, turn a campaign position into an effective regulation or change the terms of a cloud contract. The incoming Congress begins in January 2027. IT planning needs to connect each political development to the particular decision it could affect.

AI policy: expect a different negotiation

The White House’s March 2026 national AI legislative framework already asks Congress to act on issues including children’s protection, AI-enabled scams and data-center permitting. It is a legislative framework, rather than a law enacted by its announcement. The election changes the political setting in which those requests will be considered.

Our assessment is that IT leaders should prepare for contested choices rather than assume either an immediate crackdown or a clear path to deregulation. A bill’s scope, the organizations it covers, implementation dates and eventual enforcement provisions will determine the work. An expectation of more oversight is not yet a requirement to run a particular algorithm audit.

Consider an organization deciding whether to embed an AI assistant in a customer workflow. It can distinguish commitments that are expensive to reverse from choices it can revisit. A contract that allows the provider to change data use, retention or model access deserves attention now because the organization must operate under that contract regardless of which proposal becomes law. A speculative compliance feature can remain a separately priced option until there is a defined requirement. This is a procurement approach, not a prediction about the final legislation.

Data centers: the project and jurisdiction matter

Texas shows why attributing every infrastructure constraint to the result would be misleading. On September 21, Governor Greg Abbott directed the Texas Commission on Environmental Quality, the state’s environmental regulator, to halt permits sought by data centers pending an audit by grid operator ERCOT. His announcement also required completion of ERCOT and Texas Water Development Board audits before permits could proceed. Those instructions came before the election.

Gina Hinojosa’s campaign separately called for a moratorium on data-center builds until the legislature could establish protections. That statement establishes her campaign position; it does not establish a new law taking effect on election night. The scope and implementation of any subsequent action will need to be checked in the actual government documents.

For an IT team, a proposed facility and an operating cloud region are different exposures. A migration that depends on capacity from an unfinished site may be sensitive to a permit or interconnection delay. An existing service is not shown to be unavailable merely because a politician proposes restrictions on new construction.

Suppose a supplier’s expansion is central to next year’s rollout. Ask which specific facility supports the commitment, what approvals remain outstanding and what capacity is available if the opening slips. The answer may change the rollout sequence or the contract negotiation. It does not justify assuming a national cloud shortage, a price increase or a service outage from the election alone.

Cybersecurity funding: follow the decision that changes the service

The administration’s fiscal 2027 budget proposes a $707 million reduction for the Cybersecurity and Infrastructure Security Agency (CISA). A presidential budget request is evidence of the administration’s priorities, not proof that Congress enacted that reduction. The election adds another variable to the funding debate; it does not tell a security team which specific assistance or information-sharing service will be available on a particular date.

Organizations that rely on federal or state assistance should follow the actual appropriation and the relevant agency’s service announcements. For example, a team whose incident plan depends on a government contact should confirm that contact and the supported response process directly. That is worthwhile preparation without asserting that federal intelligence has disappeared or that a particular staffing cut occurred.

The near-term effect of the midterms is a change in the decisions IT teams need to watch. Treat an enacted requirement as implementation work, a pending proposal as a scenario, and a supplier’s documented project constraint as a commercial dependency. Putting all three into the same urgent compliance bucket would obscure both the work that is required today and the choices still open.

Permalink

Post-Quantum Is an Inventory Problem Before It Is a Cryptography Problem

A post-quantum plan cannot begin with, “Which new algorithm should we buy?” It begins with a less glamorous question: where, exactly, does the organization rely on public-key cryptography, and which of those dependencies can actually be changed?

NIST’s new transition report is useful partly because it makes the answer larger than an encryption setting. The change can touch applications, cryptographic libraries, hardware, certificate infrastructure, network protocols, cloud services, and the vendors that maintain them. If those dependencies are unknown, a date on a roadmap is mostly decorative.

An inventory reveals the replacement problem

Consider a business with a customer portal, a VPN, laptops that verify software updates, a hosted line-of-business application, and a device that will remain in service for years. Each may use public-key cryptography differently. Some may be updated by a vendor. Some may need a firmware update, a new certificate process, or eventual replacement. Some may already use a service whose provider controls the underlying change.

That is why a useful first inventory is not a search for every mathematical detail. It is a record of the systems that protect important information or establish trust, what information must remain confidential for a long time, who supplies each system, and what the supplier says about its upgrade or replacement path. The record will be incomplete at first. That is still more actionable than treating the organization as one undifferentiated “encryption project.”

NIST IR 8547 describes the same practical layering: applications depend on libraries and hardware; public-key infrastructure has its own issuance, validation, and revocation work; protocols may need to accommodate different key sizes or interfaces. A change at one layer does not establish that the others are ready.

Not every item has the same clock

Post-quantum planning is often discussed as if a future quantum computer will ring a bell and every organization must migrate on the same afternoon. NIST does not describe it that way. Its federal timelines apply to particular federal systems, and the report says migration details and schedules vary by use case.

The difference matters. The “harvest now, decrypt later” risk is most relevant when someone can collect encrypted information today and that information must remain confidential far into the future. A short-lived transaction, a system that can be upgraded quickly, and a device with a long replacement cycle are not necessarily the same planning problem. Digital signatures, authentication, encrypted archives, and software-update verification also create different dependencies.

This is not an argument to wait for a prediction about when quantum computers will become capable enough to break current public-key systems. NIST says no such cryptanalytically relevant machine currently exists, while also noting that cryptographic migrations can take a long time. The sensible conclusion is narrower: identify which commitments last longest and which systems move slowest before deciding what should be first.

Ask suppliers for a path, not a promise

An inventory becomes useful when it produces specific conversations. For each consequential system, ask: What public-key functions does this product or service use? Who controls the upgrade? Is a supported post-quantum or hybrid option documented? Does a dependency such as a hardware module, certificate authority, remote peer, or older device limit the change? If the product cannot support the transition, is there a replacement or retirement path?

Those questions are not a demand for every vendor to deploy a new algorithm tomorrow. They are a way to discover where a future transition depends on contracts, refresh cycles, interoperability, or a decision someone else must make. NIST notes that systems unable to support post-quantum cryptography should be identified for replacement or decommissioning in the federal context; for other organizations, the comparable task is to know whether such a dependency exists before it becomes an urgent surprise.

Make the first deliverable small enough to use

A business does not need to turn this into a speculative technology program. Start with one bounded record: the service or system, the protected information or trust function, the owner and supplier, the expected life of the data or equipment, and the documented next step. Mark unknowns plainly. A supplier’s unsupported answer is useful evidence too.

That record will not choose an algorithm or prove that a migration is complete. It does something earlier and more valuable: it turns post-quantum cryptography from a distant noun into a list of business dependencies that can be scheduled, funded, or retired on purpose.

Permalink

What If You Bought an IT Outcome Instead of IT Hours?

Imagine a future managed-IT proposal that does not sell a block of technician time or a ticket queue. It promises something narrower and more useful: a new employee can work, a routine access problem is resolved, a known security event is contained, and the business can see whether those things actually happened. That is a hypothetical purchase, not a description of a CTS service or a claim that such a complete product is available today.

Pieces of that proposition already exist. The harder question is whether the pieces amount to an IT department that can be bought as software—or whether they only make a human-run service desk more efficient.

A resolved ticket is not yet an operating outcome

Some current tools take action across the systems behind a common employee request. Moveworks, for example, describes workflows spanning service management, identity, endpoint management and monitoring for actions such as password resets and VPN troubleshooting. That is more than a chatbot finding an article. But it is vendor documentation, not an independent measure of what happens across every application, device and unusual request at a 25-person company.

The useful unit of work is also larger than the request. “I cannot connect to the VPN” may involve the account, device posture, network path, remote-access service and a change made earlier in the day. Resolving it means restoring the employee’s ability to do their work without opening an unsafe path or obscuring the cause. A system that closes the ticket quickly but leaves the user unable to reach a critical application has completed an administrative event, not the business outcome.

Software needs a way to check its own work

That distinction is why automated validation matters. HPE’s Marvis Minis documentation describes synthetic wired and wireless client checks that test items such as DHCP, DNS, authentication and application reachability even when no user is present. Its documented scope depends on specified network equipment, software versions and subscriptions. Still, the idea is important: an operating service needs evidence that a service works, not merely evidence that a command was sent.

In a hypothetical purchase, the provider could make an outcome claim concrete: after a change, show that the designated user path works; after a network alert, show which expected path was tested and what was not. That would give a buyer something more useful than a dashboard saying that an automation ran.

Authority and recovery remain part of the product

Automation does not remove the question of who is allowed to act. NIST’s zero-trust architecture separates authentication from authorization: proving who or what is making a request does not by itself establish permission for that request. An agent that can change passwords, install software or modify access needs bounded authority just as an employee does.

The same is true when an automated security control acts quickly. Microsoft Defender for Business documents automatic containment of devices and user connections after it detects certain advanced attacks, while leaving investigation and release actions for authorized administrators. That is a useful division of labor, not evidence that a customer can dispense with review, recovery decisions or accountable ownership.

For a buyer, the practical questions are specific. Which actions can the service take without asking? Which action requires a named approver? What evidence shows that the action worked? Who can reverse it, and how does the service operate if its own control plane or an integration is unavailable? Those questions do not assume that humans are permanently indispensable. They define the evidence needed to compare a software-operated service with a people-operated one.

Buy the test, not the slogan

An IT service could become increasingly software-like without becoming universally autonomous. The relevant commercial test is not whether a demo resolves a password reset. It is whether a provider can state the supported outcomes, show the systems and permissions required, measure success across normal work, and price the exceptions honestly.

If independent evidence eventually shows that a product can deliver those outcomes across a representative business environment at lower total cost and with manageable exception work, “we need a provider because IT requires people” will be a weak reason to reject it. Until then, a buyer does not need a prediction about the end of managed services. They need a proposal that makes its operating promises observable.

Permalink

Before the First Call From the New Office

Before signing off on an office move, ask the phone provider what location an emergency call from the new office will report. A successful call to a customer answers a different question.

For a business using internet-based phone service, the move includes an emergency-location task. The National 911 Program advises owners to contact their VoIP provider when moving a phone so its physical location can be updated. A familiar number and a working internet connection should not be treated as confirmation that this has happened.

Put an address on the acceptance check

Consider a hypothetical company moving from one leased office to another. It keeps its telephone numbers. Staff can receive calls at the new desks and call customers. Those checks are useful: the business needs ordinary calling to work on opening morning.

Now give the move coordinator one additional acceptance question: what emergency location is associated with the phones at those desks? The answer might already be correct. It might require a provider update, a change by the phone-system administrator, or confirmation that automatic location detection recognizes the new network. The point of the question is to find out which applies before declaring the move finished.

This is a recommendation for organizing the handoff, not an account of a CTS customer move. The coordinator does not need to configure emergency calling personally. They need the responsible provider or administrator to explain how the installed service identifies location and what has been checked at the new premises.

Automatic location still has a setup

Some systems can determine an emergency location from the network a device connects to. Microsoft’s Teams documentation describes a service that matches a client’s network information against an administrator-defined map of emergency locations. That map can use Wi-Fi access points, switches or subnets.

This is an important qualification to the simple advice to update an address. An organization may have dynamic emergency calling rather than a single address that stays attached to a phone. The check should fit the actual system. For Teams, the documented mechanism depends on configured network and location information; automatic detection is not evidence that every new office has already been mapped.

Microsoft also documents emergency-address management separately from assigning locations to users. It notes that management and requirements vary with the phone connection option. That makes a useful distinction for the handoff: having the new address in an administration screen is one fact; establishing how the relevant users or devices obtain that location is another.

For the hypothetical move, ask for an explanation tied to the new premises and the ways people will call there. If staff use both desk phones and a laptop calling app, include both in that conversation. Do not assume that checking one device establishes the behavior of every other device or work location.

Arrange verification without an unscheduled 911 call

A sensible move record would identify the premises, the calling arrangements checked, who confirmed the emergency-location setup, and anything still unresolved. This is our proposed acceptance record, not a universal technical procedure or proof of legal compliance. Keep the provider’s instructions with it rather than inventing a test from a generic checklist.

The National 911 Program says a test call can confirm reception and location information, but it should be arranged through the local 911 center’s non-emergency number. Contact the provider about its supported checks and coordinate any emergency-service test in advance. Do not dial 911 just to obtain that non-emergency number or make an unarranged test.

Then close the office-move task with a specific result: the emergency location for the new premises has been confirmed through the applicable process, or a named part of that confirmation remains outstanding. That gives the person accepting the move something more useful than “the phones work.”

Permalink

Getting Back Into a Controller Can Erase Its Program

Attackers changed passwords and network addresses on computers used to monitor and control water systems. Utilities lost monitoring or control, and reported effects included pressure loss and flooding, according to a July 30, 2026 FBI/EPA warning. For some affected controllers, restoring access poses another problem: the recovery procedure erases the program that runs the equipment.

The agencies’ announcement covered incidents reported since July 27 by utilities in at least seven states. That is the July account, not an updated victim count. Rockwell’s recovery guidance, updated September 18, explains what restoring access requires for particular controller models. It does not tell us which procedure any individual utility used or how long its recovery took.

The program has to come from somewhere

A controller is a small industrial computer assigned to monitor equipment, control it, or both. Losing access can therefore have different consequences: operators may lose their view of a process, the ability to control equipment, or both. The FBI says the effect depended on the controller’s job and the facility’s ability to switch to manual operation.

Rockwell’s guidance addresses controllers whose passwords are unknown. For the models it covers, recovering access involves clearing the stored program and loading a retained project file. The manufacturer warns that an offline copy is needed before the reset. Once the program has been erased, the controller cannot provide the missing copy.

That creates two separate recovery tasks: regain access to the device and restore the instructions it should execute. A successful reset can complete the first task while leaving the second unfinished. The useful question about a backup is therefore whether it contains the program appropriate for the equipment now in service.

An old file may predate a legitimate change. A more recent copy could contain an unauthorized one. Those are possibilities to check, not findings about any particular utility. The FBI recommends reviewing controller project files against known-good logic and checking backups for malicious changes before deployment. A filename and a recent date cannot establish what the program will do.

The usual connection may not work

There is also a physical requirement that routine remote maintenance can leave untested. For the MicroLogix 1100, one of the controller models identified in the July warning, Rockwell specifies a serial connection for the recovery operation. That means having the appropriate cable and a compatible computer connection available. The procedure does not support Ethernet, the usual wired network connection.

This limitation belongs to that model’s documented recovery route; it is not a rule for every industrial controller. Its practical significance is straightforward: being able to reach a device over the network does not establish that the same equipment can recover it after a lockout. The qualified integrator needs to identify the applicable software, connection hardware and local access requirements for the installed model.

These are preparation checks, not instructions to reset a live process. A restoration demonstration belongs in a separately authorized, safe test arrangement. The manufacturer’s notice provides recovery and hardening guidance; it does not identify a newly disclosed software vulnerability.

Keep the process working during recovery

The agencies also recommend practicing manual operation. That serves a different purpose from blocking an intruder: it can provide a way to continue an essential process while normal monitoring or control is unavailable. Whether it is workable depends on the equipment and the people operating it. A written fallback should not be mistaken for a demonstrated capability.

For an organization that depends on a controller, a useful preparation check starts with one function: establish how it could operate safely during a control outage, locate and verify the retained program, and confirm that the appropriate recovery tools and authorized site access are available. This is our recommendation drawn from the documented dependencies, not a reconstruction of a utility’s response or a report of CTS testing.

The check should leave any missing prerequisite plainly identified. The available records may establish that a backup exists while leaving its contents unverified, or identify the correct connection while leaving the necessary hardware unavailable. Those are specific problems that can be resolved before access to the controller is lost.

Permalink

Which Apps Can Change the Controller?

An approved application presents valid credentials to a network controller, then asks for information it has no reason to use. That is a hypothetical threat scenario in NIST’s O-RAN security report: code added to an otherwise approved app could collect and export information outside its normal work. The credentials could still check out. NIST describes a possible attack, not a reported incident.

The distinction is useful when evaluating an application for a controller. Establishing which app sent a request leaves another decision: whether that app should be able to make that particular request. A product demonstration may show the useful result while leaving the extent of the access difficult to see.

When a controller accepts third-party applications, the buying conversation needs to reach the permissions of each application. Which information may it read? Which commands may it issue? How closely do those capabilities match the job it was installed to perform?

Start with the request

The report concerns the RAN Intelligent Controller in Open Radio Access Networks, a mobile-network architecture. It describes specialized applications that obtain information or influence network operation through the controller. Section 2, printed page 6, separates a request from an impersonator from an unnecessary request made by an approved component. That second possibility supplies the example above.

The enterprise comparison here is narrower than the telecom architecture: examine the authority attached to an application. Consider the following illustrative building-control purchase. An organization wants a weekly report identifying rooms whose temperatures fall outside a chosen range. The proposed reporting app reads measurements and produces that report. Changing the building’s operating schedule would be a separate capability.

For this example, the first permission review can stay close to those two actions. The buyer asks the integrator to identify the measurements the report consumes and to show whether the app can also submit schedule changes. If the integration requires a broad service account, the proposal should explain which additional actions that account permits and why they are needed.

Read access and command access have different consequences

An app that reads measurements can expose the information it receives. An app that changes a schedule can affect when equipment operates. In our hypothetical purchase, allowing both would enlarge the app’s authority beyond the stated reporting task. The organization could choose a narrower integration, accept a documented limitation, or decide that a different feature justifies additional access.

Those are decisions about a specific capability. A general statement that the application is trusted or comes from an approved supplier leaves them unanswered. The useful evidence would connect the requested permission to a named function: this information is needed to calculate the weekly report; this command is needed only if automated schedule adjustment is purchased.

The distinction also makes a demonstration more informative. In a safe test environment agreed with the integrator, the organization could ask to see the reporting workflow succeed and an out-of-scope schedule-change request refused. A screenshot of the finished report establishes the output. The second observation tests the proposed access boundary.

Revisit permissions when the job changes

Suppose the organization later adds automatic schedule adjustment. In this illustrative extension, the app’s job has changed, so write access may now have a purpose. The decision needs to specify the schedules it may alter, the conditions under which it may act, and how an operator can restore an acceptable setting. Keeping the earlier reporting-only description would conceal that expansion.

If the platform cannot separate those permissions, that constraint belongs in the purchase decision. The buyer may need an additional control, a restricted deployment, or a different integration. A demonstration can then end with a concrete record: the app can read these measurements, issue these commands, and is blocked from these other actions. Any capability the test could not establish remains an open item.

Source: NIST IR 8625, section 2, printed page 6; NIST publication record and report scope.

Permalink