---
title: "A broken election compass, a school tool nobody was meant to see, and the choice of base year | Weekly Wiseguys, 21 February 2023"
summary: "Sami Miettinen and Tere Sammallahti on Ivan Puopolo's Weekly Wiseguys, 21 February 2023, six weeks before the Finnish parliamentary election. Three stories, all of them about the same question: who decides which numbers a citizen gets to see. Helsingin Sanomat's election compass gave one question double weight, the paper conceded the error — and, the panel argues, the arithmetic still fails to reconcile after the fix. A public-broadcaster tool comparing schools was demanded taken down, and Yle kept it up. A claim that taxation has quietly fallen is taken apart on the choice of base year. The episode is tongue-in-cheek, sharp-edged and sometimes overshoots — those passages are marked too."
datePublished: 2023-02-21
dateModified: 2023-02-21
originalLang: en
section: satire
sections: ["satire","economy","society"]
authors: ["Sami Miettinen"]
tags: ["Media-vierailu","Viikon viisastelu","Weekly Wiseguys","Ivan Puopolo","Tere Sammallahti","Vaalikone","Helsingin Sanomat","Datajournalismi","Koulutus","Verotus","Julkinen talous","Sananvapaus"]
canonical: https://www.neuvottelija.com/ai/viisastelu-hsn-vaalikone-s2-tulokset-ja-verovalheet/
---
# A broken election compass, a school tool nobody was meant to see, and the choice of base year | Weekly Wiseguys, 21 February 2023

# A broken election compass, a school tool nobody was meant to see, and the choice of base year | Weekly Wiseguys, 21 February 2023

> **Summary:**
> Sami Miettinen and Tere Sammallahti on Ivan Puopolo's Weekly Wiseguys, 21 February 2023, six weeks before the Finnish parliamentary election. Three stories, all of them about the same question: who decides which numbers a citizen gets to see. Helsingin Sanomat's election compass gave one question double weight, the paper conceded the error — and, the panel argues, the arithmetic still fails to reconcile after the fix. A public-broadcaster tool comparing schools was demanded taken down, and Yle kept it up. A claim that taxation has quietly fallen is taken apart on the choice of base year. The episode is tongue-in-cheek, sharp-edged and sometimes overshoots — those passages are marked too.

This is a summary on Neuvottelija AI. The episode itself is on Ivan Puopolo’s channel and is his — [watch the original](https://www.youtube.com/watch?v=WKQGwmhNqNs). The appearance is catalogued at [neuvottelija.com/media](https://www.neuvottelija.com/media/viisastelu-hs-election-compass-s2-results-and-tax-claims/), which is the canonical home of this content.

> **A note on reading this.** Weekly Wiseguys is a tongue-in-cheek political panel, not a news broadcast. The speakers are right-of-centre commentators and politicians talking in their own names, sharpening for effect and joking as they go. This article sorts their arguments and marks who said what; it takes no position on who is right, and it does not reproduce the personal jabs. Speakers are named only where the recording itself identifies who is talking: the transcript carries no speaker labels, so uncertain passages are attributed to the panel or to "one of the guests" rather than to a name. The opinions are the speakers’ own.

---

## The panel

**Ivan Puopolo** hosts; the guests are **Tere Sammallahti** and **Sami Miettinen** — the
same line-up as the first episode in the series. It is recorded on 21 February 2023, six
weeks before the Finnish parliamentary election, and it opens with an announcement: the
programme is putting itself under an **election quarantine from 1 March to 2 April**, during
which a right-wing guest will always be paired with a left-wing one. Nobody imposed it; the
panel imposed it on itself, and it is worth reading as the frame for the episode. The panel
knows it is one-sided and says it will correct for that before polling day.

Three stories, one question: **who decides which numbers a voter gets to see.**

## 1. The election compass that counted one question twice

This is the strongest section of the episode, because the claim is checkable and it turns
out to be at least partly right.

Helsingin Sanomat's election compass places the respondent on a four-quadrant grid whose
vertical axis runs from conservative to green-liberal. One of the panellists says he filled
it in as soon as it was published and landed on the conservative side — because, he says, he
declined to give an affirmative answer to a single question about gender diversity.

A National Coalition candidate in Tampere then noticed the same thing and said so publicly.
**Helsingin Sanomat conceded the error**: that question had entered the calculation with
double weight.

The panel's argument does not stop at the concession. It is a **structural claim about the
fix**:

- Moving one single answer by one step changes the value-axis result by two to three
  percentage points — but the **overall recommendation** moved by 17 points before the fix,
  and by 26 when the answer was reversed entirely.
- *After* the fix, that same one-question difference still produced roughly a 15-point swing
  in the overall recommendation and only three points on the value axis.
- The panel's conclusion: the weighting may have been removed, but something else in the
  calculation still produces a disproportionate effect, and **the compass is still broken**.

A second checkable observation: a candidate in Pirkanmaa said he had entered exactly his own
candidate answers and got a **97 per cent** match with himself rather than 100. The same
phenomenon from the other direction: MP **Päivi Räsänen** wrote on Twitter — and the panel
reads the tweet aloud —

> *"I couldn't get myself even into my own top twenty, and nobody close to me who tried the
> compass got me onto their list either."*

A candidate who fails to match themselves is an unusually clean error signal in an election
compass: it is the one case where the correct answer is known in advance.

### Where the panel draws the line

It is worth marking **what the panel does not claim**. A suspicion surfaces in passing that
the compass might have been deliberately calibrated so that nobody lands in the conservative
corner — and the panel rejects it itself:

> *"Until some other evidence turns up, I believe it's just bad code and that's that."*

Two of them do hold on to the view that the **direction** of the error is not a neutral
matter even when the error itself is identical for every respondent. That is an opinion, not
an observation, and should be read as one.

### The demands

The panel's three concrete demands are of very different sizes, but all clearly put:

1. **Publish the compass's code and data.** The reasoning is a light verifiability argument:
   civil society could walk through the calculation quickly and the question would be settled.
2. **Take the compass down while the error is fixed.** The reasoning is timing — a compass
   steers votes while it is live, and a later correction reaches nobody.
3. **Report the correction as prominently as the compass itself.** The observation is that
   the correction went out as a small note.

### The axis the panel says is the wrong one

The idea at the centre of the episode — and the part of it that lasts longest — concerns the
vertical axis. *Conservative versus green-liberal* is the wrong opposition, the panel argues,
for two reasons: green-liberal is not the same thing as liberal, and the opposite of
conservative is progressive, not liberal. The right axis would be **liberal versus
authoritarian**:

> *"You give individuals a great deal of freedom... but at the same time you can take away
> their freedom of speech, you can take their property and you can raise the tax rate. For
> some reason none of that counts to them as an indication of illiberal behaviour."*

The supporting observation is what the compass left out: no questions on liberalising drug
policy, on legalising sex work, on plural marriage — classic liberal/conservative
discriminators — leaving a single question about gender diversity to carry the axis almost
alone. The absence of those questions is checkable; the interpretation of what the absence
means is not.

The panel also makes an honest concession that is easy to miss: **there is no such thing as
a perfect compass question.** Every question is incomplete from some direction, and the
respondent ends up playing tactics. From that they draw advice that is not polemical at all:
read the candidates' **written answers**, not just the percentage.

## 2. The school tool they were told to take down

The public broadcaster published a tool for comparing comprehensive schools using variables
assembled from open official data: leaving-certificate averages, the share of pupils
studying Finnish as a second language, and the educational and socio-economic background of
the area. Publication was followed by a demand to withdraw it — according to the panel the
demand came from politicians on both the left and the centre, and the National Agency for
Education took a public position against the tool.

The panel's main claim is **a matter of principle and is made without personalising it**:

> *"If a news organisation says: we have this relevant, accurate information, but we are not
> publishing it because we are protecting something — then it is an unbelievably poor news
> organisation. And the same goes for researchers."*

And the observation about attitude that follows from it: **the first reaction is to
forbid.** The normal course of public debate is that somebody says something, somebody else
answers, and the audience follows the arguments. Demanding that the data be removed skips
that entire stage.

### What actually follows from the data

Here the panel is more careful than the headline suggests, and the disagreement between them
is genuine.

One guest points out that the correlation between the share of second-language pupils and
the class average is **structurally unavoidable**: a child who does not speak the language of
instruction on arrival cannot be expected to reach the same leaving grades as a native
speaker. The correlation is therefore not a finding and carries no value judgement — it is
the starting position.

The other replies that this is exactly why **more variables should be added** rather than
staring at two numbers: *"you can't frame it as second-language pupils and averages, that's
it."* He also says he looked at the data and did not see an unambiguous automatic
correlation.

The most interesting part is what they would want to *ask* of the data — and this is the
episode's clearest illustration that demanding the removal of a dataset also closes off the
useful questions:

- **Two schools, both with 35 per cent second-language pupils, but averages of 8.5 and
  7.5.** What is the better one doing differently? There is no answer to that without the
  data.
- **A parallel comparison with pupils studying Swedish as a second language** on otherwise
  similar parameters.
- Does regional variation in the use of the **home-care allowance** correlate with school
  results, parents' language skills and women's employment rate?

That third point connects to the panel's most concrete policy proposal: **abolish the
home-care allowance.** The argument has two parts — it would raise the employment rate of
women of immigrant background, which is internationally low in Finland, and it would move
children into day care earlier, where they learn Finnish. The panel says the same idea
appeared in savings calculations by the think tank Libera. The claim is made without a
citation, but it is an empirical claim rather than an opinion, and it is one of the few
places in the episode where a solution is offered rather than a fault identified.

The panel also makes a broader observation that has nothing to do with immigration: if both
parents are unemployed throughout a child's upbringing, the same social problems accumulate
for that child regardless of background. **The mechanism is identical for native Finns.**
They say this out loud, and it is worth marking, because it separates the argument from an
ethnic explanation.

They also call for Danish-style labour-market reform: weaker dismissal protection, an end to
the universal applicability of collective agreements, a shorter earnings-related
unemployment benefit. That is a straightforward political position, not an observation.

### Where the episode overshoots

One guest tells a taxi-driver anecdote: the driver had twice told him that a Finnish wife
would not do, because Finnish women do not cook. From this he draws the conclusion that
**law does not change attitude** and that cultural background affects women's employment.

The same guest concedes the limitation himself — *"probably no reason to generalise"* — and
that is worth repeating here: one recurring conversation with one driver is not a dataset.
The claim that culture raises the threshold to employment may be true or false; this
anecdote settles it in neither direction. The episode has just spent twenty minutes
demanding open data and comparable numbers — and here it leans on what one person said in a
taxi.

### Credit given to the other side

The most surprising moment in the episode: Yle, the programme's standing target, **gets
credit**. It did not remove the data and did not alter it under pressure. The panel says so
directly and jokes about how it feels to praise them.

With it comes a worry expressed in general terms and without accusing anyone: **pressure
does not take down this tool, it takes down the next one.** An individual reporter may not
withstand the storm, and at the next comparable case the newsroom will remember the last
one. The panel calls that creeping self-censorship, and it is the core worry of the episode.

## 3. Taxation and the choice of base year

The third section starts from an opinion column arguing that taxation in Finland has
"quietly" become lighter and that the tax instrument should therefore be used.

**Sami Miettinen** — he is named aloud during this section as the speaker — answers with
figures: realised tax revenue in 2021 relative to nominal GDP was about **43 per cent**,
which he says was the fifth highest in the world, past Sweden. He distinguishes this from
**forecast** tax ratios, from which he says the "lightening" claim is derived: a forecast
that may fall because of inflation is not the same thing as a realised figure.

The methodological core of the argument is the **choice of base year**, and it is the single
best observation in the episode:

> If you pick the peak-taxation year as your baseline, everything after it looks like
> lightening. If instead you take **public spending** at its 2007 level — the panel's
> recollection is 45–47 per cent of GDP against "57 or somewhere thereabouts" now — the
> question flips: how much wealth would citizens hold if spending had stayed there?

The panel's own answer is **€25 billion a year**. The figure is offered live without a
calculation behind it, and the percentages are from memory — read them as an order of
magnitude, not as statistics. The methodological point holds regardless of the numbers: the
choice of base year decides the conclusion, and that cuts both ways, including against the
panel.

A second checkable observation, made in passing: Finns received roughly **€6 billion in
dividends** in a year against roughly **€150 billion in earned income**. If half of an
€11 billion public-finance gap were to be covered from capital income, that would in practice
mean taxing every dividend in full. The figures are from memory, but the order-of-magnitude
argument is clear, and it becomes the panel's wish for the election debates: ask the
candidate where the eleven billion comes from.

### Two places where the numbers do not survive a check

In fairness, two claims in the episode should be marked as unreliable:

- **The counterfactual on the broadcaster's budget.** The panel estimates that if Yle's
  funding had stayed at the level of the turn of the millennium — their example is €50
  million, index-linked — the saving would have been €15–18 billion. The order of magnitude
  does not work: Yle's annual funding has been in the hundreds of millions, so a cumulative
  difference over twenty-odd years falls clearly short of that. The direction of the argument
  may be what they mean; the number is not.
- **The basis of the broadcasting tax.** The episode guesses that the corporate broadcasting
  tax is levied on turnover and starts at €200,000. The basis is taxable income, not
  turnover, and the amount is capped at a few thousand euros. The speakers hesitate at the
  point themselves, but the error stays on the tape.

### Rhetoric that should be read as rhetoric

The section closes with a claim the panel itself marks as a guess: that left-wing politicians
are on average less wealthy than right-wing ones and therefore lose their sense of scale once
they get access to public money.

> *"I have no statistics to throw at this, but I have a strong gut feeling..."*

That is how it has to be left: a feeling, not an observation. The same goes for the panel's
prediction of what one particular party would do about censorship if it were in charge. That
is a claim about other people's intentions and cannot be verified, and it should be kept
separate from the episode's checkable claims.

One point in the same stretch is genuine and checkable, though: **the exceptions that
suspended budget discipline are coming back.** The EU deficit rules were paused because of
the crises, and the panel's observation is that they land on the next government's desk
whoever forms it. So too the claim the panel treats as the most important thing said about
the economy all spring: the pandemic and the war explain part of the public deficit, but not
its scale — the roughly €12 billion hole in the finance ministry's winter forecast.

## How it ends

The episode closes on news that broke during recording: the minister for economic affairs
says the platform company has given him information about the messaging account he claims was
hijacked, but will not say what the information is. The panel had gone through the affair at
length in the previous episode, and here it is left open — one of the speakers says he has
begun to doubt his earlier reading, because the minister keeps pursuing it systematically and
a liar would be taking a considerable risk in doing so.

It is the smallest moment in the episode and possibly the best one: the panel returns to its
own earlier conclusion and weakens it itself.

## How the episode runs

- **00:00** — An election quarantine from 1 March: balancing the panel
- **01:37** — Guests they would like during the quarantine
- **02:25** — A running gag with the audience
- **03:11** — The three items on today's agenda
- **03:59** — Helsingin Sanomat publishes its election compass
- **04:49** — One question on gender diversity pushes the answer to the conservative side
- **05:36** — The paper concedes the error: the question was counted twice
- **06:32** — The arithmetic that still doesn't reconcile after the fix
- **08:06** — The correction came as a small note
- **08:56** — The error is identical for everyone — but which way does it point
- **09:45** — The wrong dichotomy: green-liberal versus conservative
- **10:31** — The real opposite pair would be liberal versus authoritarian
- **11:41** — What the compass left out: drugs, sex work, plural marriage
- **13:21** — Why the written answers matter more than the score
- **14:08** — A market-economy question settled back in the 1980s
- **15:45** — An interlude
- **16:31** — The candidate who couldn't match himself: 97 per cent
- **17:17** — The demand: take the compass down and publish the code
- **18:04** — Päivi Räsänen's tweet: she couldn't get herself into her own top twenty
- **18:51** — A calibration bias, or just bad code?
- **20:26** — Many voters decide on the compass twenty minutes before the polls close
- **21:13** — Yle's school tool: data journalism from an open register
- **22:01** — The demand to take the data down
- **22:47** — The politician's reaction and the researcher's duty
- **23:36** — When a news organisation withholds information it knows is true
- **24:22** — A correlation that is built into the situation
- **25:09** — More money has gone in, results have gone down
- **26:02** — The first reaction is to forbid
- **26:56** — The questions the data would raise: Swedish-as-a-second-language, peer schools
- **27:42** — "Protecting children" — from what, exactly
- **28:31** — Two schools, the same share of language learners, different averages
- **29:25** — Home-care allowance, employment and language skills
- **30:11** — The solutions go missing and the labels come out
- **30:59** — Employment through the parents — the same mechanism for everyone
- **31:54** — The Danish labour-market reforms
- **32:41** — The taxi-driver anecdote, and the limit of generalising from it
- **34:20** — Credit where due: Yle kept the data up
- **35:07** — Pressure on reporters and creeping self-censorship
- **35:52** — The Charlie Hebdo comparison
- **36:38** — Taxation: the claim that the burden has quietly fallen
- **37:25** — 43 per cent, and where that ranks internationally
- **38:15** — The choice of base year decides the conclusion
- **39:06** — What if public spending were at its 2007 level?
- **39:52** — The mantra that public spending can never be trimmed
- **40:45** — Yle competing on YouTube with tax money
- **41:34** — The details of the broadcasting tax get muddled on air
- **42:22** — Campaign talk about the economy
- **43:14** — "There's money" and how a sense of scale gets lost
- **44:01** — Dividends, six billion, and an eleven-billion gap
- **44:49** — The kind of questions asked in election debates
- **45:39** — A slogan: money to the citizens
- **46:26** — The EU deficit rules land on the next government
- **47:21** — The pandemic and the war do not on their own explain the hole
- **48:18** — A minister's allegedly hijacked messaging account
- **50:08** — Sign-off and a security seminar