Then I Actually Drove the Scirocco

There is a dangerous period in every opinion.

It begins after you have collected enough information to feel confident.

And ends when you actually try the thing.

In June 2012, I had an opinion about the Volkswagen Scirocco.

A strong one.

The new Opel Astra Cosmo, I wrote, had a 1.4-liter engine producing 140 horsepower and cost around 52,000 lira.

The Scirocco had 122 horsepower and cost around 63,000.

Therefore, naturally, anyone paying more money for less horsepower had some explaining to do.

The conclusion seemed obvious.

The numbers were there.

What else was there to discuss?

I even mocked one of the features people mentioned.

Dual-zone climate control.

One side could blow at 18 degrees while the other side blew at 17.

Why would a human being need this?

Was one shoulder living in a different climate zone?

Was the passenger experiencing an independent weather system?

From the perspective of a spreadsheet, I was winning the argument magnificently.

Then I added an edit.

Today I got in one.

Nice car.

That edit may be one of the shortest records of intellectual defeat I have ever written.

And I like it very much.

Not because the Scirocco was objectively superior.

Not because my original comparison was completely wrong.

Because the edit captures something engineers, reviewers, consumers, and opinionated people everywhere repeatedly forget:

Specifications are not experience.

This should be obvious.

It rarely is.

We love numbers because numbers are clean.

Horsepower.

Price.

Screen size.

Battery capacity.

Weight.

Memory.

Clock speed.

Dimensions.

Fuel consumption.

Benchmark scores.

Numbers let us compare things before touching them.

That is useful.

Without measurable properties, purchasing becomes pure impression.

But numbers also create false confidence.

If one car has more horsepower and costs less, the conclusion begins to feel mathematical.

Except a car is not horsepower divided by price.

A car is a system.

Steering.

Seat position.

Visibility.

Suspension.

Transmission.

Noise.

Materials.

Pedal response.

Cabin shape.

How the door closes.

How the dashboard sits around you.

How the car behaves when you turn.

How it makes an ordinary drive feel.

Many of these things can be measured.

Almost none are captured by one number.

The same problem appears in technology constantly.

This laptop has the faster processor.

That phone has more RAM.

This camera has more megapixels.

That monitor has a higher refresh rate.

This product wins.

Then you use both.

And discover that the “worse” one is nicer.

The keyboard feels better.

The interface is cleaner.

The battery behaves more predictably.

The screen is easier on the eyes.

The software gets out of the way.

The product feels coherent.

Coherence is difficult to put in a specification table.

Yet it is often what we experience most.

I think that was what happened with the Scirocco.

Before sitting in one, I was comparing attributes.

After sitting in one, I was evaluating a car.

Those are different activities.

The first can be done from a distance.

The second requires contact.

This is why reviews are difficult.

A reviewer wants to justify an impression.

“I liked it” sounds weak.

So we search for numbers to support the feeling.

Sometimes the feeling came first.

The explanation came later.

This is not necessarily dishonest.

Human judgment works that way.

We experience the whole thing, then try to decompose the experience.

Comfort.

Quality.

Response.

Design.

Value.

The categories are attempts to explain something already felt.

The mistake is pretending the categories completely generate the feeling.

They do not.

The original entry is funny to me because I had constructed a perfectly reasonable case.

The Astra was cheaper.

More powerful on paper.

I preferred its styling.

I saw little value in certain features.

Everything aligned.

Then one ride forced me to append:

Nice car.

That is a good outcome.

Being wrong after direct experience is much healthier than protecting a bad opinion.

But protecting the opinion is tempting.

Once we argue for something publicly, the argument becomes attached to identity.

Changing your mind begins to feel like losing.

So people move the goalposts.

Yes, but…

Of course it feels nice, but…

The numbers still say…

The other one is still better value…

Sometimes those qualifications are valid.

Sometimes they are just defensive engineering around the ego.

I could have done that.

Instead, the edit was almost embarrassingly simple.

I got in.

It was nice.

That does not erase the price difference.

It does not make the specifications irrelevant.

It adds information.

That is how opinions should work.

New evidence arrives.

Update.

No ceremony required.

This is surprisingly difficult in public life.

People often think consistency means never changing a conclusion.

Real consistency should mean applying the same reasoning process when the evidence changes.

If your rule is:

Judge based on available evidence,

then changing your mind after better evidence is not inconsistency.

It is the most consistent thing you can do.

Engineering teaches this constantly.

The simulation says one thing.

The prototype says another.

Which do you trust?

The model matters.

But if the real system behaves differently, the model needs work.

You do not tell the hardware:

Please behave according to my spreadsheet.

Reality has no obligation to preserve your argument.

Cars are excellent teachers of this because they combine measurable engineering with subjective experience.

Two vehicles can be close in power and completely different to drive.

A slower car can feel faster.

A heavier car can feel more stable.

A cheaper car can feel better assembled.

A more expensive car can disappoint.

The relationship between technical merit and user experience is not linear.

Software is the same.

A technically elegant tool can be miserable to use.

A less elegant one can make people productive.

Programmers are especially vulnerable to dismissing usability.

We admire architecture.

Users admire finishing their work.

The product has to survive both evaluations.

This does not mean subjective experience should dominate everything.

Feelings can also mislead.

A car can feel fast and be inefficient.

A user interface can feel simple because it hides dangerous complexity.

A product can create a premium impression while being unreliable.

The correct lesson is not:

Ignore specifications.

It is:

Specifications are evidence, not verdicts.

That distinction has helped me far beyond buying things.

It applies to people too.

We build specification sheets for people constantly.

Education.

Job title.

Age.

Salary.

Where they live.

What they drive.

Which university.

How many years of experience.

Which programming languages.

We create a profile.

Then we predict the person.

Sometimes accurately.

Often badly.

A résumé is not a colleague.

A dating profile is not a relationship.

A reputation is not a conversation.

A title is not competence.

Contact changes things.

Someone impressive on paper may be impossible to work with.

Someone ordinary on paper may be extraordinary in practice.

This does not mean background information is useless.

It means the system is larger than the variables we selected.

That is exactly what I had forgotten with the car.

I had chosen variables.

Horsepower.

Price.

Appearance.

Climate control.

Then acted as if the variables were the object.

They were not.

There is another reason the old entry makes me smile.

It shows how confidently younger versions of ourselves explain things they have not experienced.

This is not unique to youth.

But youth has more untested confidence available.

You read.

Compare.

Conclude.

Experience eventually introduces friction.

Jobs are not what job descriptions suggested.

Cities are not what photographs suggested.

People are not what first impressions suggested.

Technologies are not what benchmarks suggested.

Relationships are not what theories suggested.

Cars are not what horsepower figures suggested.

Slowly, your opinions become less clean.

This can feel like losing certainty.

It is actually gaining resolution.

Reality becomes more complicated because you are seeing more of it.

There was a time when I thought complication was evidence that I had not understood the problem well enough.

Sometimes it is.

Other times the problem really is complicated.

Maturity includes learning not to simplify past usefulness.

The Scirocco did not become cheap because I liked it.

The Astra did not become bad.

One test did not settle every possible comparison.

But my previous certainty became impossible to defend.

That was enough.

In technical work, this is a valuable habit:

separate your conclusion from your ego.

The conclusion is allowed to fail.

That is what testing is for.

If every failed prediction feels like a personal failure, you will unconsciously avoid good tests.

You will choose evidence that protects you.

That is dangerous.

A good engineer should enjoy discovering that reality disagrees early.

Better in a prototype than after production.

Better in a test drive than after purchase.

Better in a small experiment than after a large commitment.

The cost of changing your mind rises over time.

So seek contact early.

Try the software.

Drive the car.

Talk to the person.

Build the prototype.

Run the benchmark yourself.

Visit the place.

Do not confuse descriptions with experience.

There is also a lesson here about value.

Why would someone pay more for fewer horsepower?

Because people do not buy horsepower.

They buy cars.

The same misunderstanding causes endless arguments.

Why pay more for this laptop?

Why buy that phone?

Why choose that hotel?

Why eat there?

Why live in that neighborhood?

Why use that tool?

If we reduce the object to the metric we personally care about, other people’s choices look irrational.

Sometimes they are irrational.

Often they optimize for a different objective function.

Comfort.

Design.

Reliability.

Status.

Convenience.

Familiarity.

Resale value.

Service network.

Noise.

Emotional attachment.

Taste.

The fact that I do not value a feature does not make the feature valueless.

My old climate-control joke is a good example.

One-degree temperature independence sounded ridiculous to me.

Maybe it still does.

But the broader feature represented passenger control.

Some people care.

Some do not.

Products exist in the space between measurable capability and individual preference.

There is no universal equation for value.

This is why arguing about consumer products can become strangely personal.

People are not only defending objects.

They are defending priorities.

My purchase says these things matter.

Your criticism sounds like criticism of the things I value.

Then everyone becomes a lawyer for a machine.

Probably unnecessary.

A better response is often:

Not for me.

That sentence saves enormous amounts of time.

The Scirocco may still not have been the car I would have bought.

That is not the interesting part.

The interesting part is that direct experience forced an update in seconds.

The specifications had given me a conclusion.

The car gave me doubt.

Doubt won.

Good.

That is what experience is supposed to do.

I sometimes wish all strong opinions came with a built-in test drive.

Political opinions.

Career advice.

Technology arguments.

Life plans.

You would make the confident statement.

Then reality would hand you the keys.

Go ahead.

Try it.

See what you think afterward.

Many opinions would survive.

Many would need an edit.

Mine did.

The original entry spent several lines explaining why paying extra for the Scirocco made no sense.

Then came the edit:

Today I got in one.

Nice car.

No long apology.

No attempt to hide the contradiction.

Just new information.

I think that may be the best part of the story.

Not the car.

The edit.

We should all be willing to leave more edits under our old certainties.