Permits & Development Accountability

projects · The Log is quarterly, not live, and covers large-scale projects only. The publisher states: "All information that is presented here is accurate to the best of our knowledge at the time of publication. All information stated is subject to revision or retraction at any time." Counts verified at the 2026-07-30 refresh: Building Permit Granted 33, Zoning Permit Granted or As of Right 28, Complete 5, Pre-Permitting 4, Approved PUD/Master Plan Development Remaining 3, Design Review 1, Permitting 1.

homes (permitted residential units) · The field counts PERMITTED units, which is not the same as occupied units. A blank is not an unpublished value: the publisher defines it as "Number of permitted residential units included in project including new construction and those in rehabbed structures. Blank value indicates nonresidential project." Residential Units is populated on 46 of the 75 projects; the other 29 are nonresidential by that definition. Verified at the 2026-07-30 refresh.

percent of role-disclosed filings, with both raw counts shown · The field records what the applicant declared on the form. It is not independently verified by the City, and Authorized Agent does not say whom the agent acts for. Addition/Alteration figures are as of that dataset's 2026-08-05 refresh; Demolition as of 2026-08-16.

percent of all filings, with raw counts · A blank means the field was not captured. It cannot be read as Property Owner and cannot be read as Authorized Agent, and it is the reason T06-M03 states a share of role-disclosed filings rather than of all filings.

permits, as one fraction per role — four fractions, never a combined disclosure score · Four separate fractions and no index. Rolling them into one number would hide that one role is disclosed on every record and another on none, which is precisely the weakest link a single score conceals. Verified at the dataset's 2026-08-16 refresh.

percent of recorded reviews carrying a reviewer and a date · The permit datasets publish departmental review as booleans — that a review was required or performed — with no reviewing body, no review date, no outcome and no conditions. The flags are also thinly populated: on the 362 new-construction permits, community development review is recorded on 4 records, zoning plan review on 3, electrical plan review on 20 and plumbing plan review on 21. So the review dimension is real and per-department, and both its content and most of its coverage are unpublished.

percent of permits, with counts · No Cambridge permit dataset publishes an owner name. Addition/Alteration publishes an Owner-Occupied flag — true on 5,705 records, false on 1,021, and not recorded on 7,446 of 14,172 at that dataset's 2026-08-05 refresh — which says whether the applicant claims to occupy the property, not who owns it. Ownership sits in the Assessing Department's property database and the Registry of Deeds, on a different update cycle.

percent of certificates, with raw counts · A joinable pair says the paperwork closed. It says nothing about unit counts, affordability or conditions. 24 certificates carry no permit number and cannot be traced backwards from this dataset alone. Verified at the 2026-08-16 refresh.

percent of permit records, with raw counts · 8,102 records carry no parcel number and cannot be placed on a parcel timeline at all. Address text is present on many of them, but matching an address string to a parcel is a geocoding step with its own error rate, and this site does not perform it silently. Verified at the 2026-08-16 refresh.

records per status value · Three claims about the same field are in circulation and none has been reconciled, so all three stay visible. The consequence is concrete: no approval, denial or withdrawal rate can be computed from this dataset, and any product publishing one is not reading this field.

date range · A coverage window is the first thing a time series needs and the last thing most dashboards check. Any trend drawn across 2013 to 2020 from this dataset would be drawn across records that are not in it, and the site does not quietly start an axis at 2013.

projects · 47 of 75 projects publish a building permit number: 33 at Building Permit Granted, 5 at Complete, and 9 at Zoning Permit Granted or As of Right. The published field note says the number appears only at the first two of those stages, so it does not reliably predict which projects carry a key — and 28 projects carry none.

percent of filings, with counts · Structurally not computable from the building-permit datasets. Inspectional Services publishes the limitation itself on both New Construction and Addition/Alteration: "Only completed applications are listed in this dataset." The published records are therefore a survivorship sample of the filings. Separately, the Permit Finder's Status field carries exactly two values citywide — Complete 88,146 and Active 37,069 at the 2026-08-16 refresh — with no denied, withdrawn or revoked state anywhere in it.

days — median and range, per deciding body and per relief type · Two of the three bodies publish no case-level dataset. For the third, a workflow date IS published: 754 of the 755 Board of Zoning Appeal Request records carry a Decision Stamped By City Clerk date. That is the date a signed decision was filed, not the date the board voted, and it carries no outcome — so a figure built from it would measure time to filing, not time to decision. The dataset's decision-deadline columns are statutory: "Date the decision must be filed with City Clerk." A deadline is the date by which a body must act, never the date on which it did.

dwelling units · The field is populated on all 327 records and sums to 849.75 — a fractional number of homes. That is a data-entry artefact in the source, and it is published here as 849.75 rather than rounded to 850, because rounding it away would hide the only evidence a reader has that the column is imperfect.