Hacker News Reader

Comment by andylynch | Hacker News Reader

Immutability of information in lab notebooks is also important for auditability. Pen in a bound paper notebook works well for this. Missing pages or erasures are a huge red flag for inspectors and rightly so.

View on HN

This is the case in some fields, particularly life sciences where it's very difficult to reproduce results, but in others it's not important at all. I have colleagues who have to have their labbook pages signed by the group leader each day, and others that can even take their labbooks with them when they leave their jobs because no one minds/cares.

Indeed, and different fields operate under different regulatory / legal environments. When my parents were working as research scientists, they followed a rigid set of guidelines that were considered to be the gold standard at the time for use of notebooks in patent disputes.

Labs in regulated environments such as medical or pharma research have their own rules today, and an industry of vendors who sell enterprise level software for managing data in a compliant fashion.

I was working at a company on the day that US patent law was reconciled with the rest of the world, and the rules for priority changed. That was the last day we were required to keep a lab notebook.

Today, I mainly keep records so I can figure out what I did later on, and the best single tool that works for me is Jupyter.

Wouldn’t git with signed commits work just as well? You’d have to find a good app that saved stuff as text (not Word, but probably not LaTeX either since that takes time to get the formatting right), but once you figured that out, you’d have a full history of all changes made.

Yup - that's probably a good equivalent.

Signed commits still let you falsify your own work after the fact. You need an external trusted service, a digital notary essentially, that would provide a signed record that you submitted a commit on a certain date with certain set of ancestor commits. Each time you added an entry, you would submit the hashes of all the entries in the book to the digital notary and get a signed record that you submitted those commits on that date. Inability to produce this record would be similar to having erasures or pages torn out in a lab book.

a public blockchain would do for the digital notary? or hash posted on twitter?

You can do this in a pretty cheap and low-tech way by publishing the hash of your current state somewhere that will keep a permanent record. E.g. the Bitcoin blockchain or a classified ad in a local newspaper.

This. True lab notebooks are not just personal notes, they are a legal record. Not every page of every lab notebook will matter, but it absolutely happens and is one of the many real use cases for keeping a lab notebook. If you invent something or discover something of significance, you will want this. Paper notebooks have pages stitched together and ink that can be run through chemical analysis to see if all the text on the page was written in a single session by the same pen. Forensic examiners have about 100 years of experience assessing paper documents. Digital systems can provide legal grade archival support but a sweet front end editor with copy paste and markdown support does not imply anything about traceability of the contents.

A VCS like Fossil at least makes the parent commits un-fungible, so you would only need to sign the latest commit, and not every commit every single time. Git is amazing in it's flexibility, but it's not necessarily always the right VCS solution.

Otherwise I agree with your answer about Git.

Yep! This is already done and for exactly this purpose; I believe the common term is "trusted timestamping" [1].

[1] https://en.wikipedia.org/wiki/Trusted_timestamping

That seems difficult unless it is completely automatic. Also, you will have to convince auditors that your implementation works and that your users are using it correctly and that everything has been committed correctly and in the correct order. So while it might be obvious to you that this works "just as well" it is far from obvious to others.

You wouldn't need to sign the commits, you just need to give the latest hash to a trusted (by the authorities) third party, and that will validate the existence of all history up to that point at that time, so ideally at the end of each day, you give your SHA-1 HEAD to the authority.

This is basically what the block chain does, except that instead of having somebody designated as special and trusted, they have the signers perform proof of work.

I remember that i read about a lawsuite were someone tweeted the hash of a document and proofed that way that this documenta existed at that point. I believe it was about a patent infringment and they even brought a twitter employee in as a witness to testify aboht the imutability of tweeta and the timestamp.

Cant find the article (hash is not a good search phrase when it comes to Findung an article about twitter).

Interesting, what about the fact people can just not write something in it?

Reply on news.ycombinator.com