Field OP
Terms of ServicePrivacy NoticeAcceptable Use PolicyData Processing TermsLabelling TermsInferred Trait ScheduleSub-processors

Every characteristic Field OP can hold about a person — which it computes, which an organization has switched on, and which it merely permits.

This is a draft. It is not in force and it is not legal advice.

This is a draft written by Field OP’s engineers so that counsel has something concrete to react to. It has had no legal review, it is not legal advice, and it is not in force. Nobody has been asked to accept it and nobody should be. The bracketed markers left in the text are the open questions, kept in place rather than smoothed over.

Field OP — Inferred Trait Schedule

Version: 1.0-draft Published at: /terms/trait-schedule and /terms/trait-schedule/1.0-draft Effective: not in force. This Schedule is engineering's draft, no lawyer has reviewed it, and the page carrying it says so before the text.

This Schedule forms part of the Field OP Terms of Service and is referred to in section 9.1. It lists every characteristic the platform can hold about a person — which it computes, which your organization has switched on, and which it merely permits — as at this version.

We may add to or change this Schedule on notice to your administrators. A change that adds a new category of characteristic — rather than a further characteristic within a category already named in section 9.1 of the Terms of Service — is a change to those terms and is handled under section 16.3.

Everything in section 9 of the Terms of Service and in the Labelling Terms applies to every characteristic in this Schedule: each is an estimate and never a fact; each is stored apart from the person's record and marked as inferred; a trait the person states about themselves outranks the guess; no message is ever selected from a guess and nobody is included in or excluded from contact on one; and reading any of them requires the Restricted Data grant.

The platform refuses to write a kind your organization has not enabled, and it refuses it where the label would be written rather than in the interface. That is the control, and it is narrower than this Schedule: what may be written for you is the list your administrators hold on Admin → Labelling.

The database's own list of permitted kinds is wider than the characteristics this Schedule used to name, and the rest of it is set out at the foot of this Schedule. Nothing computes those, nothing enables them, and no label of any of them exists — but a characteristic the platform could store and this Schedule did not name would be one your acceptance did not cover, so they are listed.

Which characteristics your organization starts with. An organization that signs up from anywhere in Canada but Québec is created with labelling on and with all six of the characteristics below enabled. An organization that gives a Québec address is created with labelling off and no characteristic enabled at all. Either can be changed by an administrator at any time.


Computed from a person's family name

These come from the name model. It is a guess from a family name and nothing else — not from anything the person told you, and not from anything they did.

CharacteristicWhat is estimatedStatus
LanguageThe language a person is most likely to read or speakOn by default outside Québec. Computed
Ethno-cultural communityThe community a family name is most commonly associated withOn by default outside Québec. Computed
ReligionA religion, derived by applying a published assumption about a community to a person. The least reliable estimate the platform produces; usually returns a spread rather than a single answerOn by default outside Québec. Computed, and it declines to name an answer far more often than the other two

Computed from records your organization already holds

CharacteristicWhat is estimatedStatus
Residency stabilityWhether a person has likely moved. A rules estimate drawn from what a door reported, the person's own address history, and a membership that ended. It reads no nameOn by default outside Québec, and computed today. Off for a Québec organization, with everything else

Declared but not computed

These appear in the platform's vocabulary and in the database's list of permitted kinds, and — this is the part the earlier wording missed — they are enabled for a new organization outside Québec along with the four above. Enabled is not the same as computed: nothing computes them, and no label of either kind exists. They are listed because you are entitled to the whole vocabulary rather than the part that is currently in use.

CharacteristicStatus
ContactabilityEnabled by default outside Québec. Nothing computes it
GenderEnabled by default outside Québec. Nothing computes it

If either becomes computed, this Schedule changes first and your administrators are given notice. Because they are already enabled, the change that would start producing them is a change to the platform and not a change to your settings — which is exactly why they are named here.


Permitted by the database, and nothing more

Six further kinds sit in the database's list of permitted kinds, inherited from the platform's wider vocabulary of scores. Nothing computes them, nothing enables them, no organization has them switched on, and no label of any of them exists.

CharacteristicStatus
SupportPermitted by the database. Not enabled, not computed
TurnoutPermitted by the database. Not enabled, not computed
PersuadabilityPermitted by the database. Not enabled, not computed
VolunteerPermitted by the database. Not enabled, not computed
SignPermitted by the database. Not enabled, not computed
DonorPermitted by the database. Not enabled, not computed

With the six characteristics above them, that is twelve kinds the database will accept. This Schedule is complete when it names all twelve, and it is only protective if it is complete: a characteristic the platform could store and this Schedule did not name would be one your acceptance did not cover.


How to read the accuracy figures

The accuracy of the name model is reported in the product, ordered worst-community-first, and your administrators can see it before and after switching the feature on. It never reports a single averaged figure, because an average hides both ends.

Those figures are measured on a synthetic test population built to be classifiable. A real population is read less accurately, not more. On a real campaign the report cannot break down by community at all, because nobody knows anybody's actual community; only a person's own correction can measure it, one at a time, after the fact.

Section 9.5 of the Terms of Service sets out what this means in practice.


Field OP is a product of Gestalt Communications Inc.

How to cite this

A record of acceptance names a document, a version and a hash of the exact wording shown. A version is never edited — a correction is the next version — so these three identify one set of words for ever, and check:legal fails the build if the words behind a published hash change.

Document
trait-schedule
Version
1.0-draft
Wording (SHA-256)
cb7df7696ca151ba98b28382266b8b3040215379466cf09c7bb3469e3658b6f8
Length
6,511 bytes
Master handoff
§15.15

This is the only version so far. It lives at its own address, /terms/trait-schedule/1.0-draft, and will stay there when a later one replaces it here.

Field OP is operated by Gestalt Communications Inc., Toronto. Questions about a campaign’s use of your information go to that campaign; questions about the platform go to info@fieldop.ca.