Count Votes with Meek's Method RCV / STV

Please confirm the following suggested candidate IDs for each candidate.

After making any desired adjustments, click / tap the 'OK' button to accept the listed candidate IDs. Click / tap the 'Cancel' button to cancel the "Read File" operation.

OK Cancel Help

One or more candidate IDs have errors or other anomalies that are likely to cause later difficulties.

Do you want to:

• Generate ballots and other data and deal with any issues later, or
• Change candidate IDs first.

Candidate IDs with problems are marked with a strong red border.

Fill in the data for your election following the patterns of the sample input. The input uses the JSON format . Then click on / tap the "Tabulate" button to get results.

Topics

• Candidate IDs
• Shortcut for a list of candidate IDs
• nbrSeatsToFill
• maxRankingLevels
• candidates
• names
• tieBreaker
• excluded and protected candidates
• ballots
• options
• randomSeed
• JSON format errors
• Browser-based tabulations
• Results
• Read File

Candidate IDs identify candidates and are typically shortened or abbreviated candidate names. Candidate IDs make it easier to specify how candidates are ranked on ballots.

A candidate ID is not allowed to begin with ":" or be equal to "#" or "".

The candidate IDs should not include spaces, other white-space characters, or any of the following characters: ' [ : " ] ' (not including the enclosing single quotes). It is recommended that any punctuation in candidate IDs be limited to hyphens, underscores, and periods and that the first character be an alphabetic letter. It can be helpful if the the candidate IDs are all the same length.

For example, a candidate named Thomas Jefferson might be given a candidate ID of "TJ", "TJ1", "TJeff", "ThJe", "ThomJ", or "ThomJeff".

A shortcut for specifying a list of candidate IDs uses a delimiter-first string. The candidate IDs are all put in a single string, separated by a delimiter character which is also added as the first character The following delimiter-first strings are equivalent to each other and will be converted to an array of strings:

• " Jill Juan Jane Jack"
• ",Jill,Juan,,Jane,Jack"
• "|Jill|Juan||Jane|Jack"
• "+Jill+Juan++Jane+Jack"

The equivalent standard JSON array of strings is:

• ["Jill", "Juan", "", "Jane", "Jack"]

However an array of strings can involve typing and reading a lot of punctuation. It is usually simpler and easier to use the delimiter-first strings instead.

The delimiter can be any character, but using the space character or a punctuation character is recommended.

Using this shortcut, the empty string, "", indicates an empty array of strings, [ ]. Similarly, a one-character delimiter-first string, such as ":", indicates the array with just one empty string, [""].

This shortcut can be used to specify a list of candidate IDs for the "candidates", "tieBreaker", "excluded", and "protected" items, and for the rankings of a ballot. It can also be used for specifying round-by-round values for the "alternative_defeats" options sub-item. For example, use:

• " Y N N Y N" instead of
• ["Y", "N", "N", "Y", "N"]

The "nbrSeatsToFill" item indicates the number of winners that will be elected by the tabulation.

Fewer candidates might be elected if there are not enough available candidates, i.e. candidates that are not excluded.

The "maxRankingLevels" item indicates how many candidates a voter is allowed to rank. The number specified must be at least 3. A null value, the letters "null" without the quote marks, can be specified to indicate that there was no limit on how many candidates a voter could rank.

This item is used to determine whether exhausted votes are reported as abstentions. If a ballot is exhausted and the voter had the opportunity to rank additional candidates, but didn't, any portion of that ballot's vote which is not counting for a candidate is reported in the ":Abstentions" category.

The "candidates" item lists the candidate IDs for each candidate. The list can be given as an array of strings or as a delimiter-first string. The order in which candidate IDs are listed is not important.

The "names" item provides full candidate names, keyed by candidate ID. The item value should be a JSON object. For example, a names item for two candidates could be specified as:

where "TJ" and "SA" were declared as candidate IDs in the candidates item.

The "tieBreaker" item is a list of candidates used to break ties for being defeated. Candidates are typically listed in a random order. The list of candidates can be given as an array of strings or as a delimiter-first string.

Among tied candidates, the candidate listed first in "tieBreaker" will be defeated. It is a tabulation error if a candidate is involved in a tie to be defeated but is not listed in the tieBreaker.

The "excluded" item lists the candidate IDs of any candidates which are excluded (a.k.a. withdrawn) from the tabulation. Excluded candidates are considered defeated before the first round of tabulation starts. The list of excluded candidate IDs can be given as an array of strings or as a delimiter-first string. The order in which candidates are listed does not matter.

The "protected" item lists the candidate IDs of any candidates which are protected from defeat. Protected candidates as a result are guaranteed of being elected. The list of protected candidates can be given as an array of strings or as a delimiter-first string. The order in which candidate IDs are listed does not matter.

A re-tabulation of ballots to fill one or more vacancies can be accomplished by:

• Excluding any candidate who is not eligible or is unwilling to serve.
• Protecting any incumbents who will continue to serve.
• Setting the "nbrSeatsToFill" equal to the number of vacancies being filled, plus the number of protected candidates.

The correct number of protected winners and the correct number of unprotected winners in such a re-tabulation is ensured by using separate thresholds for each class of candidates.

The "ballots" item is specified as a JSON array of ballot groups. A ballot group represents one or more ballots with the same candidate rankings. Use of ballot groups allows ballots to be specified in a summarized and typically more compact format.

Each ballot group is specified as a JSON array of two items: the number of ballots in the group and the candidate rankings for those ballots. The candidate rankings can be given as an array of strings or as a delimiter-first string. The candidate rankings list the first-choice candidate first, the second-choice candidate second, etc.

The following are equivalent ways of specifying a ballot group that represents 7 ballots with the same rankings:

• [7, ["A", "B", "C", "D"]],
• [7, " A B C D"],
• [7, "|A|B|C|D"],

If the ballot group represents just a single ballot, only the ballot rankings have to be provided. The following are equivalent ways of specifying a ballot group that represents one ballot:

• [1, ["A", "B", "C", "D"]],
• [1, " A B C D"],
• [1, "-A-B-C-D"],
• ["A", "B", "C", "D"],
• " A B C D",
• "-A-B-C-D",

Besides the candidate IDs, there are two additional values that can be used to specify the a ranking in the rankings for a ballot group:

• The empty string, "", represents a ranking level without a ranked candidate. Such unused rankings can be omitted, since their presence does not change how votes are counted. In a delimiter-first string, the empty string is represented by two consecutive delimiter characters and by a delimiter character that is the last character of the string.
• The value "#", a single pound sign character (a.k.a. hash sign, number sign, and octothorpe character) represents an overvoted ranking. An overvoted ranking occurs when more than one candidate was ranked at the same ranking level.

Note that the last ballot group in a list of ballot groups is not followed by a comma. Instead it is followed by the right square bracket that closes the list of ballot groups.

If the ballots and ballot rankings are available in some other format and there are very many of them, it may be worthwhile to format the data using another tool, for example using a spreadsheet.

The "options" item identifies various alternative ways to perform the tabulation. If an options sub-item is not specified, a default value, the value first listed below, takes effect. Using the defaults by specifying an empty options value, { }, is usually a good way to run a tabulation, unless there are specific reasons to use other values.

The recognized options sub-items and their allowed values are:

• "alternative_defeats" specifies whether more than one candidate is allowed to be defeated in a round, subject to various conditions that generally are designed to ensure that multiple defeats will not change who is elected.

  Alternative defeats are called multiple simulateous defeats in the reference rule. They are also known as batch defeats and defeats with deferred surplus distribution.

  Valid values are:

  • "N", which disallows checking for alternative defeats in all rounds.
  • "Y", which allows checking for alternative defeats in any round.
  • an array or a delimiter-first string of "Y" and "N" values in any combination, one for each round of the tabulation.

• "type_of_altdefs" specifies when during a round any checking for alternative defeats is performed, if it has been allowed by the "alternative_defeats" option. Valid values are:

  • "if_no_new_electeds", which will check for alternative defeats in every iteration, provided that no candidates have been elected in the round, but regardless of whether the round requires another iteration. Specifying "if_no_new_electeds" produces the combined effects of both "per_reference_rule" and "before_single_defeats".
  • "per_reference_rule", which will follow the optional provisions for "Multiple simultaneous defeats" in the reference rule. That will check for alternative defeats every iteration, provided that no candidates have been elected in the round and that the round will require at least one more iteration. This can mean that there is no checking for alternative defeats in a round which finishes in just one iteration, for example in a first round that does not elect anyone.
  • "before_single_defeats", which will check for alternative defeats only at the end of a round in which no candidates have been elected, and just before a regular single defeat would be required.

• "always_count_votes" specifies whether votes should always be counted in a first round, even if the winners can be determined without any rounds of tabulation, i.e. when the number on non-excluded candidates is equal to or less than the number of seats to fill. Valid values are:

  • true, which requires that votes to be tabulated for at least the first round.
  • false, which follows the reference rule and avoids all rounds of tabulations if the number of non-excluded candidates is equal to or less than the number of seats to fill.

• "ballots_tree" specifies whether to use some optimizations in how ballots are internally represented during a tabulation. Using these optimizations can help tabulations run faster.

  All three options are designed to produce the exact same tabulation results. Vote totals for each iteration and each round do not change. Valid values are:

  • "dynamic" is the fastest option typically. It maintains a tree with rankings maximally summarized and with only the rankings that could be used in a round. The tree is adjusted between rounds.
  • "static" is typically a close second in speed to the dynamic option. It builds the full tree before starting any tabulation rounds and does not adjust the tree between rounds.
  • "none" is the slowest, but simplest option. It counts votes for each ballot group separately.

For example, specifying an empty options value, { }, is equivalent to specifying the options item with all of its default values:

  ,"options": {
    "alternative_defeats": "N",
    "type_of_altdefs": "if_no_new_electeds",
    "always_count_votes": true,
    "ballots_tree": "dynamic"
  }
    

The "randomSeed" item provides a way to create a randomized tieBreaker when reading and converting from a file that is in an Election Buddy ballot format. The value can be any string or number. The default value is the string "undefined".

The conversion process generates a tieBreaker value using SHA-256 hashes of the randomSeed value and the names of candidates. If those values do not change, repeated conversions will produce the same tieBreaker. The relative order of listed candidates in the generated tieBreaker does not depend on the presence or naming of any write-in candidates.

Once a tieBreaker has been generated with a randomized randomSeed value, it is recommended that all subsequent tabulations use that tieBreaker value.

If the input data is not in a valid JSON format, for example maybe it has a missing comma or an extra comma somewhere, trying to perform a tabulation will result in a JSON parse syntax error being reported. The error message will indicate the row and column where an error was detected. To find the corresponding place in the input data, it may be helpful to copy and paste the input data into a text editor that reports the row and column of the cursor. The place where the input data needs to be corrected may be at an earlier point in the input data, but the spot indicated by the error message is a good place to start looking for problems.

This web page does not send any of your election data to the web server or anywhere else on the network. The web server is only used to provide the files needed to display the web page. The vote tabulations are performed entirely within your browser on your computer.

One important consequence is that this web page does not save any of your data. If you need to save your input data or the tabulation results, you should do that by copying and pasting the data to another program, for example a text editor or a spreadsheet, and then using that program to save your data.

This web page provides an option to perform tabulations in the background. Background tabulations are still done completely within your browser. However when longer running tabulations are done in the background, you will see progress messages, and you also are able to cancel the tabulation without leaving the web page.

This web page was designed for you to be able to save the web page to an HTML file on your computer and then later access the saved HTML file to perform tabulations. Due to limitations imposed by modern browsers, when a saved HTML file is accessed directly by the browser, you will not be able to tabulate in the background.

A successful tabulation shows the results in three levels of detail:

• Who was elected, both as a JSON array of candidate IDs and as a plain text list of candidate names (or candidate ID in lieu of name).
• The status of every candidate in a tabular format:
  • The candidate's name (or ID in lieu of a name).
  • The candidate's ID.
  • Whether the candidate was elected or defeated.
  • The round in which the candidate was elected or defeated.
  • The number of votes the candidate had when elected or defeated.
  • The candidate's keep value at the end of the last round.
  • If the candidate was protected or excluded, a string indicating that.
  Rows for elected candidates are highlighted with light green.
• A round-by-round tally of votes for each candidate and other statistics in a tabular format with the following features:
  • The vote total for the round when a candidate was elected is highlighted in green. Subsequent vote totals for the elected candidate are highlighted in a lighter green.
  • The vote total for the round when a candidate was defeated is highlighted in red. For subsequent rounds, the defeated candand does not have vote totals.
  • Rows of candidates in groups of four are otherwise highlighted in alternating shades of light blue. This highlighting only intended to help you visually follow a row across a wide table and does not indicate any substantive differences among the candidates.

All of this data is also shown in a JSON format.

Both the tabular data and the JSON data can be copied and pasted into a spreadsheet for other report formatting, analysis, or graphical presentations.

After pasting the JSON data into a spreadsheet, you may need to use a text-to-columns command to split the data into a tabular format. Use of a text-to-columns command should recognize as separators: spaces, commas, and all of the characters in ' [ : " ] ' (not including the enclosing single quotes). It should also merge delimiters. If the spreadsheet program, such as LibreCalc, offers an option to recognize special numbers, that option should be activated.

In Microsoft Excel, it may be necessary to first use "Find & Select" / "Replace" to first remove the square brackets and the double quote characters in the Tally data before using the text-to-columns command.

A candidate's keep value for the last round is a value between zero and one which indicates what portion of votes received by the candidate were kept and counted for the candidate, rather than being passed on as surplus to less preferred candidates. For candidates elected before the last round, the keep value will typically be a value less than one but more than zero. For example if a candidate is reported to have 300.0 votes in the last round and a last-round keep value of 0.75, that tells us that in the last round the candidate received about 400 votes, but about 100 of those votes were transferred as surplus from the candidate.

For both the status and tally portions of the results, candidates are listed in an order that reflects the progress of round-by-round tabulation results:

• Elected candidates are listed before defeated candidates.
• Elected candidates are listed in the order they were elected.
• Defeated candidates are listed in the reverse order that they were defeated.

Among the benefits of using that ordering is that the tabular formats show the progression of electing candidates at the top and defeating candidates at the bottom. Also, the candidates who are still competing to be elected or defeated in the last round are listed close to each other in the middle.

Clicking / tapping on the "Read File" button will allow you to select a file on your computer that is used to populate the input area.

The file can be a text file with JSON data or it can be a file using the format for Election Buddy ballot files (a vote_by_vote.csv file).

If the file is not recognized as an Election Buddy ballot file, the current text in the input area will be completely replaced by the file contents. This allows you to easily start with data that you earlier saved or with data that is separately created and maintained.

To save data from the input area, copy and paste it to a plain text editor and save the data from there. This web page is not able to write data to files. Note that the input area will likely have a simple undo capability for changes made by typing, but it may not be able to undo the effects of reading a file. So first save any data that you might want to reuse.

If the file is recognized as being formatted as an Election Buddy ballot file, the current text in the input area must be valid JSON or must first be manually deleted. Clicking on the "Show Sample Input" is a quick way to fill the input area with valid JSON data.

The results of the conversion will replace any values for the candidates, tieBreaker, ballots, excluded, protected, and names items. All other values will be kept, but might be rearranged in a different order or with different formatting.

The tieBreaker that is generated by the conversion depends in part on the randomSeed item. The randomSeed value should be set to a randomized value.

Candidate names will be stored in the input area as a "names" item. These names are not used by the tabulation, but are available to be copy-and-pasted to other results reporting capabilities.

As part of the data conversion, suggested candidate IDs are generated for all of the candidates. Those candidate IDs are subject to your review before the new data is created and put in the input area. It is possible that some of the suggested candidate IDs might not be unique and would need to be adjusted during your review. During the review, any candidate IDs that are in error or are otherwise problematic are flagged with a red border.

Candidate IDs can also be manually entered in the fourth line of the EB-format ballot file. For listed candidates, put the candidate ID in the candidate's tab-separated column. For write-in candidates, put the candidate ID in subsequent columns in the order in which write-in candidates are encountered in the file's ballot lines.

The Election Buddy ballot files can be read by most spreadsheet programs as a CSV (comma-separated-value) file, but with settings to recognize only the the tab character as a value separator and not using any quoting character. A single-quote (') or double-quote (") character can be used as the quoting character if no value (particularly a candidate name) begins with that character. If any changes to the file are saved from the spreadsheet program, similar settings for tab separators and not quoting values should be used.