An informal LinkedIn post described asking software developers whether they would want their children to pursue software development.

The reported answer was no.

It is a striking result—and, without the original post, exact question, sample, date, response options, and method, it is not a finding Violet can verify. It should not become evidence that developers as a group believe software development has no future.

But it raises a useful communication question:

Were the respondents forecasting the future of software, or answering from the moment in which the question reached them?

A developer answering during layoffs, while using AI in daily work, and after repeated predictions about coding automation does not imagine the future from nowhere. The question arrives inside an existing field of experience, risk, conversation, and attention.

The same thing happens when:

  • a customer rates a product immediately after a failed payment;
  • employees receive an optimistic AI announcement after a restructuring;
  • voters hear a message about growth after markets have fallen and borrowing costs have risen;
  • patients receive reassurance after guidance has changed;
  • or residents are asked to evaluate a public project after a prior concern appeared to go unanswered.

In each case, the response may look as though it is about the focal question or message alone. It also inherits the conditions around it.

That does not make the response irrational, biased, or irrelevant. The surrounding moment may contain exactly the evidence the sender failed to consider.

For research, product, communication, and change teams, misreading that response produces the wrong intervention. A team may rewrite a survey when timing is the problem, add product education when a failed journey has damaged trust, or repeat a leadership message when the unresolved issue is the source’s credibility. The practical task is to identify what the response measured before deciding what to change.

A future-facing question does not turn someone into a forecaster

“Would you want your child to become a software developer?” sounds like a prediction question. It may actually combine several different judgements:

  1. What do you expect software work to look like in the future?
  2. How secure does your own career feel today?
  3. Would the work provide a life you want for someone you protect?
  4. How much uncertainty are you willing to accept on your child’s behalf?
  5. How do you feel about your employer, industry, or work at this moment?

Those are not interchangeable.

The words “your child” also change the stake. They ask the respondent to move from evaluating an occupation in general to imagining a consequential decision for someone they care about. That mental simulation can operate as an induction cue: an upstream prompt that may temporarily foreground protection, responsibility, loss, or opportunity before the person answers.

The resulting “no” may therefore contain:

  • a forecast about labour demand;
  • a current evaluation of the work;
  • a protective preference;
  • frustration with present conditions;
  • or some combination of all four.

If the question does not separate them, the answer cannot separate them for us.

Survey researchers have long warned that wording, response options, question order, and preceding questions can alter what respondents consider when they answer. Pew Research Center’s questionnaire guidance gives practical examples of wording and order effects and recommends pretesting new questions.

Research on projection bias also suggests that people can overestimate how closely their future preferences will resemble their current preferences. That is relevant here, but it should not be used as a universal explanation. A current concern can be both state-conditioned and factually justified.

The evidence about software careers is mixed

The current evidence does not support a simple “software development is finished” conclusion. It also does not support telling developers that their concerns are imaginary.

The 2025 Stack Overflow Developer Survey found that 64% of respondents did not see AI as a threat to their current job, down from 68% the previous year. On the same survey’s AI section, 52% said AI tools or agents had positively affected their productivity, while 46% distrusted AI-tool accuracy and 33% trusted it.

Hope, use, distrust, and threat can coexist in the same population—and sometimes in the same person.

There are also reasons for concern. A Stanford Digital Economy Lab research note reports declining employment among early-career workers in several AI-exposed occupations, including software development. The authors describe an association and uneven labour-market pattern, not proof that AI alone caused every decline.

At the same time, the U.S. Bureau of Labor Statistics still projects software-developer employment to grow 15.8% from 2024 to 2034. The Computing Research Association’s 2025 Taulbee data show another mixed picture: new computer-science majors declined 12.9% in its longitudinal cohort while bachelor’s degree production reached a record high.

These sources measure different populations, periods, and outcomes. They should not be forced into a single verdict.

A developer can reasonably observe a difficult entry-level market, accelerating automation, and changes to daily work while the occupation as a whole is still projected to grow. A “no” can be a legitimate response to present risk without being a reliable twenty-year forecast.

History changes the meaning of the answer

Career sentiment has moved with economic conditions before.

After the dot-com collapse, Computing Research Association reporting documented a steep multi-year decline in new computer-science majors. That did not prove computing had no future. It showed that the path into the field looked different after a visible industry shock.

Nor is it always safe to dismiss present conditions as temporary noise. NBER’s summary of research on graduating in a recession reports sizeable initial earnings losses that can take years to fade. The moment can change both perception and actual opportunity.

The right interpretation is therefore not:

People are merely projecting the present, so their answer can be ignored.

It is:

The answer combines the question with the conditions under which the receiver is evaluating it. We need to identify what the response is measuring before acting on it.

Political messages inherit credibility conditions

The same principle applies outside surveys.

On 23 September 2022, the United Kingdom’s government announced a Growth Plan that included substantial tax cuts without an accompanying Office for Budget Responsibility forecast. The House of Commons Library’s account says financial markets reacted negatively and that most economists identified a UK-specific component in rising government borrowing costs that was largely due to the mini-budget.

Prime Minister Liz Truss continued to argue for growth. In her 14 October press-conference remarks, she grounded that ambition in better jobs, new businesses, and improved living standards.

Those goals can sound attractive in the abstract.

But by then, the message did not arrive in the abstract. It arrived after market turmoil, higher borrowing costs, policy reversals, and damage to the government’s credibility. Receivers could hear “growth” through an immediate question:

Can this source deliver growth without creating more instability?

The context did not merely distract people from the message. It supplied evidence about the source and the proposed action.

This is why it would be wrong to say the speech caused the market reaction. The more useful observation is that the speech inherited a changed credibility environment, and the same words could no longer do the same work.

Violet’s Truss Growth Plan evidence case separates the documented sequence from the framework’s interpretation.

Product feedback inherits the journey before the question

Consider a product team that asks:

How satisfied are you with the platform?

The prompt appears to request an overall product evaluation. But imagine it appears:

  • immediately after an unexpected charge;
  • after the user has failed twice to complete a task;
  • during an outage;
  • after a successful resolution by a support agent;
  • or on a dashboard the user visits only when something is wrong.

The timing and placement help determine which experience is most available when the person answers.

That does not invalidate the feedback. A failed payment may reveal a material problem. But the team should not automatically treat the score as a stable evaluation of every part of the product.

This distinction matters for:

  • customer-satisfaction surveys;
  • cancellation questions;
  • onboarding feedback;
  • pricing and packaging research;
  • campaign testing;
  • feature-concept surveys;
  • employee pulse checks;
  • and public consultation.

If the organization does not preserve the path that preceded the response, it may keep the answer and discard the information needed to interpret it.

The message is focal; the encounter is larger

Violet separates several levers that are often collapsed into “context.” Why Good Messages Fail explains the larger receiver-aware model, and the Violet white paper develops its research boundaries.

Receiver background

What the person actually brings: prior knowledge, identity, trust, experience, values, constraints, and memory.

Operative state

The temporary standpoint active at the moment of interpretation: for example, protecting a child, recovering from a service failure, evaluating a personal loss, or looking for evidence of institutional competence.

Contextual field

The external information surrounding the focal message: location, people present, timing, institution, visibility, device, norms, risk, workflow, and action constraints.

Induction cue

An optional designed prompt immediately before the focal information: a question, image, scenario, comparison, role prompt, recalled experience, or earlier interface step.

Focal message

The question, announcement, interface, speech, or campaign itself. Violet examines its experienced Spotlite, Substance, Source, Strain, and Stake relative to the receiver and the transition being asked of them.

“The moment” is useful public shorthand for the combined encounter. It is not a new Violet construct and it should not erase the distinctions above.

Seven questions for interpreting a response

Before treating a response as a verdict on a product, policy, campaign, or future, ask:

1. What exactly was the receiver asked to judge?

Separate a present evaluation, future forecast, recommendation, personal choice, and protective decision.

2. What happened immediately before the question or message?

Record the page, event, preceding questions, interaction, news, error, or conversation. Do not assume the focal item began the experience.

3. What was materially at stake?

“Would this career grow?” and “Would you recommend this life to your child?” carry different stakes.

4. Which interpretation did the wording make easiest?

Identify the role, comparison, memory, risk, or goal the prompt may have foregrounded. Treat the resulting operative state as a hypothesis unless measured.

5. Does the surrounding context contain real evidence?

A market crash, service failure, layoff, safety incident, or policy reversal is not merely a cognitive bias. It may be relevant substance.

6. What rival explanation fits the same answer?

A low product score could reflect poor support, an unexpected price, survey placement, response selection, or genuine product rejection. Keep these alternatives available.

7. What comparison would distinguish them?

Change one factor where possible:

  • ask the question at different points in the journey;
  • separate present satisfaction from future expectation;
  • compare “this career” with “your child’s career”;
  • randomize question order;
  • segment by direct exposure to the event;
  • or repeat the measure after the acute moment has passed.

The purpose is not to remove context until a preferred answer appears. It is to learn which conditions change the response and which concerns remain.

Design for interpretation, not just collection

Organizations often optimize the focal item:

  • rewrite the question;
  • shorten the announcement;
  • improve the headline;
  • add evidence;
  • or make the call to action more visible.

Those changes can help. They are incomplete if the item inherits a journey, event, or credibility problem that the new copy cannot repair.

A better design records and tests the encounter:

Receiver background + contextual field + any induction cue + focal message → attention and interpretation → observable response

This changes the operating question from:

Why did they misunderstand our message?

to:

What did our message inherit, what did the response actually measure, and which part can we responsibly change?

Sometimes the answer will be a better question or message. Sometimes it will be different timing, a repaired product path, stronger evidence, a more credible source, or acknowledgement of a real loss. Sometimes the receiver understood perfectly and rejected the offer.

Knowing the difference is more valuable than collecting another unexplained score.

If one consequential message or survey prompt is producing a reaction your team cannot interpret, send one message for a free Violet audit. If the problem spans a product journey, campaign, organizational change, or public process, discuss a larger receiver-aware engagement.