For a laboratory head: the laboratory as a whole#
You look after a shared drive that many people have written to for years. No one person knows all of it. You may want to see the laboratory as a whole: what each person’s folder holds, which lines of work run through the years, where raw data was never followed by an analysis, and what a departing member would leave behind.
Ingence compares people, projects and periods inside a laboratory. The comparison is drawn on your machine, from your files, and nothing is sent anywhere (Privacy and your data).
Reading the whole drive#
Make one project whose root is the laboratory’s shared drive, with the project kept on your own disk (Read someone else’s folder shows how to start with no file opened). Then open the Dashboard. Its first view, the Years view, shows the laboratory by time, and each row can be opened to a finer scale.
People. Where each member has a folder directly below the root, the first view has one row per person, with a bar for the experiments in each year. Ingence reads that level as a person from the folder’s position, never from a name. The setting is
themes.person_level. It is 1 by default, and empty means nobody. Check that the first level of your drive really is people. If it is years or projects, those rows are not people, and you should change the setting.Lines of work. Under a person, experiments that share words in their names are drawn together as a line of work. A second view, Overlap, shows how steady each line is when the lines are drawn again and which words two lines share. A line of work is a view that Ingence recomputes whenever it is read. It is not a conclusion, and you can give a line a name of your own, which is kept under your name.
Experiments. Each is a folder that a rule recognized as one. Its color says how much of its raw data was followed by an analysis, and you can switch to a color that marks the experiments where a person has said the analysis is somewhere else.
Periods. The axis runs across the years. Scroll on it to zoom, and drag it to move.
If the laboratory keeps more than one drive, make a project for each. Another project opens at an address of its own, and the first keeps running.
What is missing#
The No analysis found list shows experiments that have raw data and little or no analysis among the
files read. An experiment is judged only once at least five raw acquisitions were found in it, and it
is listed when fewer than half of them were followed by an analysis and it holds fewer than five
analysis files and three results tables of its own (the settings are
completeness.judged_from_acquisitions, completeness.listed_below_share,
completeness.own_analysis_files and completeness.own_result_tables). Each comes with a reason where
one can be given: that at least half of the rest of its line of work was analyzed and this one was not
(“forgotten”), or that under a fifth of the line was analyzed (“a line that stopped”). Where lines of
work could not be drawn, or the rest of the line is between a fifth and a half analyzed, it says “no
reason given”. The wording matters. It means found in what was read. A missing analysis
may be on another drive, and a missing raw file may be in a place Ingence did not look. A person can say
where an analysis is, and the next run leaves that experiment off the list. A count is always shown
beside its total, such as “3 of 8 experiments”, and a share is never shown where nothing was found to
count.
Beyond that, Ingence reports copies that are identical byte for byte, programs installed inside data folders (set aside, so they do not read as experiments), links that point at nothing, and folders that have moved.
What it does not do#
It does not rate people. It counts what it found in files. What a count means for a person is yours to judge, and a quiet row can mean a good deal besides inactivity.
It does not know who did what. It reads the position of a folder, not authorship. A file’s author lives in your folder structure and your own records, not in Ingence’s conclusions.
It has no scale above a single project yet. For now a laboratory is one project, or several projects opened side by side.
It does not change anything on the drive. It reads, and writes only into its project and a small settings folder for each user’s account.
Using it with your people#
Say what you are reading, and ask. Ingence reads names, sizes and dates, and, if you allow it, the
insides of files. A member’s folder is theirs. Tell people what you will read and why, and keep
personal folders out of the reading by adding them to scan.exclude_globs in the project’s settings
file. A list written there replaces the default, so start from the default in the
settings reference.
Decisions are signed. Every answer is added to an append-only log under the name set on that computer, and an earlier answer is never overwritten. The Dashboard has no sign-in, so anyone using the computer while it runs can answer. Run it on your own computer, not on a shared one.
Share the questions, not the folder. Export carries the questions to a colleague as one file. When their answers come back, Import asks who wrote them and records them under that name. The choices are you and the people who have already answered in this project. The project itself holds names and short extracts from your files, so treat it as you treat the drive.
When a member leaves, reading their folder before it is archived shows what it holds, what repeats, and which experiments have no analysis among the files, so you can ask them before they go.
Where to go next#
How Ingence thinks explains the four inputs behind every conclusion. Teaching Ingence your words is how the laboratory’s own vocabulary gets in, and Writing a pack is how it is carried from project to project. What the evaluation says is what has and has not been measured.