The October elections will take place without the new ISSV

A paper ballot paper, an electronic candidate list and a cyber audit belong in different parts of the electoral process. October voting can take place according to the law even if the new agenda system is postponed. A valid candidate list, in turn, says nothing by itself about the strength of competition in a municipality, and a Chamber model does not describe the election of a mayor or a senator.

What we know: Act No. 108/2026 Coll. postponed the use of selected parts of electoral legislation linked to ISSV to elections held after 31 December 2027. The Ministry of the Interior states expressly that the municipal and Senate elections of 9 and 10 October 2026 will take place according to the existing procedures. On 26 August the CZSO published a continuously updated register of candidates and lists.^[3][4][6]

What we do not know: The public sources used in the article do not contain an independent final audit of ISSV readiness, nor complete data on the costs of postponement for individual offices. From the postponement itself it therefore cannot be inferred that the system was dangerous, nor that the existing procedures are without risk.

1. The ballot box and the agenda system are not the same

Imagine an August meeting at the ministry. The election official wants paper and a data mailbox, because he knows them from previous elections and can stop a defective list on them. The system supplier wants live ISSV operation, because Act No. 88/2024 Coll. counted on it and postponement looks like an admission of inability. The cyber specialist wants postponement, because widespread testing, training and emergency procedures for an outage are not finished. The mayor of a small municipality wants calm: his secretary has no capacity to learn a new interface in the weeks when candidate lists are being filed. A commentator wants a “digital state”, because he regards paper as backwardness. A political scientist wants to warn against reading the mood of the Chamber from 22 552 municipal lists.

At first glance this is a dispute about modernity. In reality each person is describing a different problem. The official compares a known procedure and judicial review. The supplier the timetable of the act. The cyber specialist resilience against attack and outage. The mayor the capacity of one office. The commentator a visible symbol. The political scientist the electoral system.

Each participant in the meeting is judging a different part of the operation, so the mere date in the Collection does not prove the readiness of the whole chain. Under the label “digital elections” equality and secrecy of the vote, candidate registration, the staffing of commissions, audit, training, a fallback procedure and an error in a fixed statutory deadline are mixed together.

First it is necessary to determine which part of election administration is being judged. Constitutional guarantees, the registration agenda, technical resilience and political commentary use different criteria.

Elections are not software. Software is not a constitutional guarantee. And postponement is not proof that the vote is in danger.

— Jiný Kontext

2. How the administration of the October elections is assembled

When a citizen receives an envelope and ballot papers on Friday afternoon, it is easy to form the impression that “elections work”. The result, however, arises in a chain. Each layer can raise quality, lower it, or change the type of error. The same vote can be constitutionally in order and correctly exhausting. The same system can be usable in a laboratory test and not in the August peak of registrations.

Five layers of one vote

1. Constitutional framework: equality, secrecy, universality, judicial review and independent determination of the result.
2. The electoral act of the given type: a municipality, the Senate and the Chamber do not measure the same contest.
3. Agenda administration: filing of lists, verification of residence and age, staffing of commissions, methodology, training.
4. Technical carrier: paper, a data mailbox, a portal, the population register, transmission to the CZSO.
5. Reading of the result: statistics, commentary, a poll, and substitution of a municipality for the state.

The Constitution and the electoral acts determine what is permitted. They do not, however, carry lists to the office or assemble a commission at midnight. Act No. 88/2024 Coll. on election administration was to transfer part of this work into an information system. Act No. 108/2026 Coll., promulgated on 25 June 2026 and effective from 26 June, postponed this applicability to elections held after 31 December 2027.^[2][3][4] The Ministry of the Interior stated expressly that the October elections to municipalities and the Senate will run according to procedures known from previous elections.^[4]

This is not a theoretical dispute. The cut of layers determines where to turn when something fails. If a voter does not receive a ballot paper, polling-station logistics failed. If an office does not finish registration, the agenda failed, not the booth. If a portal falls during the deadline, the carrier failed — and it can also pull in a constitutional dispute about who was allowed to stand. The sentence “the elections failed” is the beginning of a diagnosis, not its end.

Illustrative diagram: The same vote can, in a different layer, with a different carrier and a different reading of the result, mean a different dispute. Comparing the booth with the portal is the same error as comparing an engine with a timetable.

3. Which errors in election administration can still be corrected

The benefit of digitalisation cannot be separated from the consequences of a specific error. A typing error in an agent’s form can be corrected within the deadline, but an outage of the portal on the last day of filing can hit the right to stand. An incorrectly recorded commission member can be replaced more easily than a registration of lists that has been challenged across the board.

The Venice Commission, in the Code of Good Practice in Electoral Matters, proceeds from the point that elections are to be organised by impartial bodies and that rules are to be stable, intelligible and known in time.^[11] OSCE/ODIHR in the handbook on observing election administration translates the same into operational language: administration is not only polling day, it is the whole cycle from registration to results.^[12] It does not follow that paper is a virtue. It follows that a new carrier is judged by the type of failure, not by a date in a contract.

This also changes the question of postponement. We do not ask only whether the software “already knows a form”. We ask which system will fail in a way that the state will catch within the deadline. A creative team may tolerate ten unusable ideas. Election administration cannot tolerate a single unannounced outage on a day when the law counts hours. A slower agenda with a clear trail may better bear the guarantee of the vote than a timetable winner without a way back.

The government commissioner for digitalisation, Lukáš Klučka, put it without ornament in June 2026: launch the system on time and risk problems that could undermine the course of the elections, or postpone it. The ministry chose the second variant.^[5] This is a justification of one specific proposal of one institution, not proof that every IT postponement is a virtue. It does, however, show that the price of an error here is not aesthetic. It is constitutional.

4. What registration numbers say — and what they do not

Numbers from a closed registration procedure are needed because they replace part of impressions with a count. The problem arises at the moment when one number begins to represent the whole of politics. On 26 August 2026 the Czech Statistical Office published the registers of candidates and lists for the elections of 9 and 10 October 2026. The registers are still updated until polling day according to the instructions of registration offices; this is not an election result.^[6][7]

As of the date of publication the CZSO stated:

• 22 552 candidate lists were registered for municipal councils.
• Voting is to take place in 14 553 permanent polling districts; in 5 021 of them Senate elections concurrently.
• In 1 842 municipalities voters will choose candidates of only one electoral party.
• In 3 municipalities the municipal elections will not take place, because a list was not filed or the number of candidates was insufficient.
• 154 candidates registered for the Senate (120 men, 34 women). Czech Television added that this is the lowest number since 1998; the CZSO itself does not present the comparison with 1998 in the cited registration report as a main figure, so this is a journalistic reading of a series, not a separate statement of the office.^[6][8]

Each figure contains a hidden decision. A list is not the same as a mandate. One list in the municipality of Kateřinice in the Vsetín area, where the CZSO recorded 30 lists of one candidate each, is not the same as 24 lists in Prague with 1 060 candidates.^[6] Senate constituency 27 Praha 1 with nine candidates is not the same as constituency 48 Rychnov nad Kněžnou with two.^[6][8] The number of 1 842 municipalities with a single party does not measure “the end of democracy”. It measures in how many municipalities a contest of more lists did not arise this time. It does not follow that elections there are invalid. It follows that the role of voting is different: confirmation of representation, not a choice between camps.

Five facts without which a registration number is only a headline

• The exact type of election, the date registration closed, and whether it is a state as of an update.
• The unit: list, candidate, municipality, polling district, constituency.
• The size of the municipality and the number of council members being elected.
• Whether it is a single list, or an absence of elections.
• What the number does not measure: turnout, preferences, a Chamber majority.

For the purchase of a political story, therefore, a small map of municipalities matters more than a foreign nationwide ranking of parties. Twenty real municipalities — with one list, with thirty lists, with a drop-out of candidates — will say more about October than one column in the Chamber.

5. Compare agendas, not the words of digitalisation

The market in state systems can be described without a false ranking of winners. The following map does not say who is modern. It shows around what work the individual layers build a product, what kind of context they bring, and which failure the buyer — here the state — should deliberately look for during a trial. The state of functions changes; what decides is the current act, the methodology and the concrete account of the agenda.

The rows are not a judgment on the quality of officials. The same state offers more than one agenda and each may behave differently during a single electoral cycle. The map is meant to prevent a different error: that an excellent portal is bought in a product that will not bear the August peak, or that postponement is read as a challenge to the vote.

Interpretive diagram: The position of the elements does not express a score. The scheme shows questions that must be answered together: the impact of an error, detection, reversibility, responsibility.

6. ISSV: an advantage that is not bought with a date in the act

Act No. 88/2024 Coll. on election administration introduced common rules of state administration of elections and with them the election administration information system. The act took effect on 1 January 2026.^[2] The system was to enable electronic filing for agents, verify data against registers and coordinate the staffing of commissions. That is practical value that a pure test of “it knows a form” does not capture. Removing friction between the registry office, the population register and the list of commission members is work, not a slogan.

Universality, however, also creates a new requirement. The user must know whether he is just using live operation, a practice mode, a fallback channel, or paper. Act No. 108/2026 Coll. postponed the applicability of key parts linked to ISSV to elections held after 31 December 2027. The Ministry of the Interior added that the system should first be used at the regular election of the president of the republic.^[3][4][5] This is the plan of one institution in one press justification, not a constitutional certainty that in 2028 nothing will fail any longer.

ISSV was not alone in the June package. The Ministry of the Interior proposed a shift also for the registry information system and for the electronic system for drafting legal rules. For all three projects it stated obstacles that the original timetables did not fully take into account.^[5] It does not follow that three postponements prove the same thing. It follows that the state this time compared a deadline with resilience, not a deadline with prestige.

What needs to be tested

The first test concerns load in a real deadline: thousands of concurrent filings, not ten practice ones. The second concerns an outage: a blackout, DDoS, an identity outage, an outage of the link to the population register. The third concerns people: whether an official at an authorised municipal office can return a defect, print a document and switch to paper without the deadline collapsing. The fourth concerns the law: whether the procedure on failure of the carrier is in a decree in advance, not in a press release.

For an individual who wants “an online candidate list at last”, integration may be decisive. For the organisation of the state the sentence “we have ISSV” is not enough. It needs to know the version, the owner of the data, the fallback channel, training, audit and the boundary beyond which a person must approve the result.

7. Postponement does not change the vote. It changes the work that prepares it

For the voter, the booth does not change on 9 and 10 October 2026. The president of the republic called the elections by Decision No. 117/2026 Coll. Voting on the territory of the Czech Republic takes place on Friday from 14.00 to 22.00 and on Saturday from 8.00 to 14.00.^[1][2] A possible second round of the Senate falls on 16 and 17 October 2026.^[8] ISSV was not entering this booth. It was not to introduce electronic voting. It was to carry administration.

The source of the comparison is the ministry leaflet for candidates and the information on postponement: candidate lists and applications for registration are not filed through ISSV.^[4][9] According to the ministry’s description, postponement does not change the way in which a voter in October 2026 hands in a paper vote. That by itself does not prove that no dispute about equality or registration can arise; it only means that these questions will be handled through the existing channels. For the state, postponement means that checks of lists, residences and commissions remain with people and with previously used procedures.

Imagine two parallel illustrations, not a report. A voter enters a school on Friday at 14:30, receives an envelope, marks her neighbours, drops the envelope in. She has the feeling that the state is holding. The secretary of an authorised office in August copies lists from paper and data mailboxes, verifies data and hunts substitutes for commissions, because the postponed software will not take these acts. Both can be right about their layer. Neither describes the whole machine.

8. Paper is not nostalgia. It is a checkable trail

Resistance to paper is often aesthetic. Paper is slow, heavy, repetitive. That is true and it is not the whole truth. A paper list with a stamp and a data mailbox with a delivery receipt give a trail that a court can read. The Supreme Administrative Court and the Constitutional Court have a long series of decisions on registration procedure and on review of elections: the dispute is about a deadline, a defect, equal treatment, not about whether the portal was modern.^[16][17]

Digital filing can improve the same trail. It can also lose it in a log that nobody owns, in an interface that falls at night, or in automatic validation that rejects a valid candidate because of an error in the register. “A PDF can be uploaded” does not mean “it is auditable, reversible and intelligible to an agent in a municipality of 400 inhabitants”.

Four-step check of the agenda

1. Existence: the filing arrived and has a time stamp.
2. Authority: the office is the one the law designated.
3. Support: a specific defect rests on a specific provision, not on an impression from a screen.
4. Scope: correction is possible within the deadline and the same for all.

The most reliable working procedure therefore separates a new carrier from new law. First a table of acts, deadlines and fallback channels arises. Only then is paper switched off. For every automatic rejection it is recorded who will correct it. If the source cannot be opened or a passage of the act cannot be found, a claim is not promoted to a fact merely because the system formulated it neatly.

9. Cyber resilience is not a stamp. It is the whole flow

The question “is the system secure?” is important, but too narrow. An organisation must know where identity runs, how long filings are kept, who has access to the population register, which third parties process hosting, and whether an outage of electricity or the network has a regime written in advance. “It passed a penetration test” — if such a test existed and were published — would still not mean “it will bear the day of the deadline”. A public audit of specific ISSV pentests as of June 2026 is missing from open sources as a separate document with a result; the ministry spoke of the need to finish widespread testing, strengthen cyber security, train users and prepare procedures for extraordinary events including a blackout.^[5]

NÚKIB, in the summary report on the state of cyber security for 2023, describes pressure on the supply chain, geopolitically motivated attacks and the need for exercises.^[13] Act No. 264/2025 Coll. on cyber security, effective from 1 November 2025, sets duties for providers of regulated services and a framework for managing cyber risks.^[14] From the public materials of this article, however, it cannot be determined which specific parts, and in what regime, apply to ISSV. A reference to the act is therefore not proof that ISSV has the same legal regime as another critical service; it only supports the requirement to describe assets, suppliers and emergency procedures.

The practical minimum is a map. It begins with the agent, continues with the office, identity, the register, hosting, the log, a backup, and ends at the place where a filing can be restored or paper accepted. Without this map a claim of resilience is a description of one part of the pipe.

The cited SAI audit conclusion from 2008 on the Integrated State Treasury System described problems of timetable and management in another state IT agenda.^[15] It is not an audit of ISSV, and one older conclusion does not authorise a claim that the same failure occurred here as well. It serves only as a reminder that a deadline, responsibility and a fallback procedure should be separate items in an audit.

10. Digitalisation of administration is not electronic voting

Part of the public debate hears “election administration information system” as “we will vote on the phone”. The election administration act and the ministry materials do not support this substitution. ISSV was to administer candidate lists and commissions. It was not to replace the ballot box.^[2][4][9]

International standards add a second distinction. The Venice Commission and ODIHR assess electronic voting as a particularly sensitive change: transparency of counting, verifiability, trust.^[11][12] The Czech postponement of ISSV does not belong in this category. It belongs in the category that ODIHR describes as election administration: registration, logistics, training, continuity. To mix these two debates is to buy fear of e-voting for a decision that did not address e-voting.

Postponement should equally not be mixed with illegitimacy. October 2026 runs according to electoral acts that courts can read: Act No. 491/2001 Coll. on elections to municipal councils and Act No. 247/1995 Coll. on elections to Parliament, as amended, and, in the scope unaffected by the postponement, also the election administration act.^[2][9][10] Postponement changes the carrier of the agenda. It does not change the right to vote.

For a European organisation that is digitalising registers, the placement of the system and the emergency regime may be more important than the last per cent in a project Gantt chart. “It can be launched on the deadline” does not mean “it is a finished electoral state”. Operation needs people, methodology, backups and a person who understands when to return to paper. A single identity outage can turn digital registration into non-registration.

11. A Chamber model does not measure a municipality. Nor the Senate

In the pre-election period another substitution of objects regularly appears. Commentary takes a nationwide model for the Chamber of Deputies and reads October from it. That is honest only when the model claims that it measures a hypothetical Chamber. It is not honest when a forecast of mayors and senators arises from it.

This is a crude misunderstanding of commensurability, not a dispute about which agency is decent.

• Chamber of Deputies: a proportional system, party brands, an electoral threshold. The act on elections to Parliament provides that only parties that exceed a set threshold are considered in the scrutiny; for a single party that is five per cent of the total valid votes.^[10]
• Senate: 27 single-member constituencies in the given cycle, a two-round majority system, specific names, a second round, different turnout. The list of constituencies for 9 and 10 October 2026 was published by the Ministry of the Interior.^[8]
• Municipalities: thousands of separate contests, accumulation of votes, associations of independents, coalitions that do not exist in a nationwide model. The political-science literature on Czech municipal elections repeatedly shows the strength of local lists and the weakness of reading a municipality as a mini-Chamber.^[19][20]

Converting Chamber preferences into a gain of Senate seats or mayors is the same error as choosing company software according to a test the firm never runs. A nationwide trend can influence a local election. It is not identical with it.

Three tests of political reading

The first is systemic: which electoral act, which unit, which round? The second is about population: who is a voter in a municipality of 400 inhabitants and who is a respondent of a nationwide telephone poll? The third is temporal: the registration state from August is not the result from October. A product that answers a day earlier need not be better if a week later nobody can find out why it answered just so.

12. After postponement, different places are checked

Before postponement it was checked whether the software would make August. After postponement it is checked whether paper and a data mailbox will bear the same agenda without a loss of equality. The act is navigation to a procedure. It is not automatic proof that the procedure will work.

One amendment can combine three true sentences into an untrue conclusion. First: ISSV is being postponed. Second: elections are being held. Third: digitalisation is a priority of the state. Each sentence works, but the synthesis “the state has given up on elections” is methodologically wrong. At other times commentary uses a high-quality secondary text on cyber matters, even though a primary ministry leaflet is available on lists being filed the old way.^[4][9]

The Constitution in Art. 102 provides that councils are elected for four years and elections are held at a time tied to the end of the term.^[18] That is why October 2026 is not an optional project milestone. It is a calendar against which software either makes it, or must get out of the way. Postponement is in this sense an admission that the calendar is harder than a Gantt chart.

Before switching on an enterprise assistant, over-shared folders are looked for in another field. Before switching on an election portal the same is looked for in another language: public test accounts, unfinished training, a missing owner of a document, substitution of search for an action that changes registration. The greatest advantage of digital administration need not be speed. It may be order in the data. The greatest weakness need not be an outage. It may be a true automatic rejection based on a figure that was true two years ago.

13. The price of a deadline is the price of a raw material, not the price of a result

The market in state IT uses several price languages at once. A project pays milestones. An act pays dates. An office pays night shifts. A voter pays with trust that he does not know he is spending. A direct comparison of “we made the deadline” without knowledge of the working procedure therefore misleads.

A cheap postponement can be economically excellent if it prevents a constitutional dispute. The same postponement can be expensive if it is repeated in two years with the same justification and nobody counts the human work of August 2026. Expensive live operation can be unnecessary for an agenda that paper will bear. It can be cheaper for a decision with a high price of error if it reduces the number of unsuccessful filings and checking work — but only when it can prove that with a test, not a slide.

The true price of the accepted result

Legislative deadline + development + tests + training + human checking + corrections + the price of a challenge + interruption of work + future migration.

The result is cheap only when it passes polling day and can be defended in court.

On the postponement proposal the ministry stated that postponement does not represent an increase in the financial costs of building the system.^[5] That is a claim about a project budget item. It does not measure night work of secretaries, court disputes about registration, or the price of losing the 2028 deadline if postponement were repeated. It is one column of the bill, not TCO.

It is therefore useful to measure the price by an accepted registration and an accepted polling-station record, not the price by a milestone. Accepted means that the act fulfilled the law, passed a check and did not cause a subsequent correction on a day when it is already impossible.

14. What to launch for which agenda

A reasonable choice does not begin with the logo of a system, but with the type of work and the checking mechanism. The following recommendations are not a ranking of products. They are a signpost to a first pilot. In each category it makes sense to choose two regimes and compare them on the same data, in the same language and with the same budget of time.

For Czech use a separate test of a small municipality is necessary. Prague City Hall will not measure whether the system preserves the meaning of a petition in a municipality where the agent is at the same time a firefighter and an accountant. The test set should contain a defective list, a coincidence of deadlines, a network outage and a text in which the correct answer is “the filing is returned for correction”.

Some teams finally need more than one regime. One for registration, a second for commissions, a third for counting, a fourth for fallback. That is not a failure of digitalisation. It is the same decision as using a different tool for accounting and a different one for printing ballot papers. The condition is that the division is intelligible, data do not pass unchecked, and people know where final authority arises.

15. A short audit that will say more than a strategy

The choice can be made without a six-year programme and without the impulse “launch because the act promised”. A short, disciplined audit before the presidential elections to which the ministry ties ISSV is enough. Its aim is not to prove that the portal can do something impressive. The aim is to find situations in which it fails, and to see whether administration can catch them.

Fourteen days from impression to a decision on live operation

1. Days 1–2: Write down 20 to 40 real acts of registration and commissions, their frequency, the price of an error and current time.
2. Days 3–4: Prepare the same inputs, mandatory points and an anonymous scoring sheet.
3. Days 5–7: Repeat every important act at least three times and keep a register of errors.
4. Days 8–10: Verify the fallback channel, permissions, export, logs, deletion and a return to paper.
5. Days 11–12: Calculate the price of an accepted act including checking and interruption of work.
6. Day 13: Stress: a blackout, a contradictory register, an incomplete filing, an old figure, a forbidden action.
7. Day 14: Decide according to the matrix, document the limits and set the date of the next review.

Before live deployment in the cycle after 2027

• Find a fallback channel that an official can use without training from the previous night.
• Determine the owner and the validity date of decisive data in the register.
• Separate search from actions that change registration.
• Introduce audit, approval and a trial group of municipalities, not only the centre.
• Verify that the agent sees from which state and version of a decision he is drawing.

If possible the assessor should not know which regime created the act. For registration he watches lawfulness, completeness and the quantity of corrections. For commissions staffing and the record. For an outage the time to paper. Each failure is classified: carrier, register, permission, training, deadline.

The pilot should also have elimination conditions. For example: a single unannounced loss of a filing, use of a forbidden figure, inability to export a record, or a situation in which the office does not know whether a list arrived. A high average score of an exercise should not override an error whose impact administration cannot accept.

The last output is not only the name of the winning supplier. It is an operating rule: what the system is used for, what not, which data it may see, what a person must approve, how the result is verified and when the decision is opened again. Acts, tariffs and threats change. Without a review date even a good postponement gradually turns into a state superstition.

16. The most expensive thing can be lost trust in the next deadline

ISSV is not to be only an electronic form. It is to connect the filing of candidate lists, data for commissions and an audit trail in a process whose statutory deadlines cannot be shifted at will.

The contest is therefore not only about the best screen. It is about electoral context. About registers, permissions, methodologies, the habits of offices. The more comfortable the ecosystem that arises, the higher the exit price can be. Automata rely on proprietary steps and people stop distinguishing their own procedure from the product in which they built it.

Integration can simplify the work of offices substantially, but it must preserve portable data and filings, a documented interface, a fallback procedure without a portal, and a regular test of readiness.

The quality of a system is not decided by the number of modules nor by a timetable in a presentation. What decides is whether it covers the real agenda, whether an error can be found and corrected in time, and whether the fallback procedure protects electoral rights.

The moment software only prints a list, the state is choosing an aid. The moment it reads the population register, decides on a candidate list and prepares a commission, it is also choosing a new structure of responsibility. And then the question is no longer: “When will we finally launch ISSV?”

Let us ask: Which error will we see in time, who can correct it, and who will bear its price — the voter, the office, or the next deadline?

— Jiný Kontext