Why You Should Waste Time Documenting Your Scientific Mistakes

This is a guest post by Jonathan Gross from Labguru.

Scientific mistakes happen. Some might call earning a graduate biology degree a comedy of errors, except the errors aren’t often funny (or cheap). As learning by trial-and-error is a key part of the graduate student journey, I recently asked the Twitterverse to share research failure stories.

One guy broke a million-dollar robot (and not just once). Incorrect primer design wasted weeks of a grad student’s time. Someone else’s biggest mistake was blindly following their professor’s advice. A Ph.D. candidate trusted a collaborator’s plasmid maps instead of performing his own sequencing. And on and on.

I’m glad that people are open about their mistakes. As in my experience, we remember these mistakes and events and they help shape our scientific careers to better or worse…

This got me thinking—many of those scientific mistakes were probably made previously by their fellow labmates.

Why doesn’t the lab learn from scientific mistakes as a team?

Scientific Mistakes

We certainly have the opportunity. At least in my lab, despite all the mistakes we heard and solved during lab meetings, we never maintained a list of “do’s” and “don’ts”. Yes there are protocols, but they capture only a fraction of the knowledge the lab has.

I left the lab long ago. Is someone else now repeating my scientific mistakes over, wasting his or her energy and sleeping hours on something that a quick note about prior experiments could resolve?

I remember losing a full month trying to troubleshoot a site directed mutagenesis experiment with a QuickChange kit. I was still new and insecure at that time… and got advice from a senior very self-confident PhD student. Just like one of the Twitterers, it wasn’t until I realized I had been following bad primer design advice that I got the experiment to work. (It would be fair to add that this student gave me loads of good advice, but the bad ones are the ones that you tend to remember.)

Did I share that knowledge so future lab members wouldn’t consume additional time troubleshooting a similar problem? No… for some inexplicable reason I kept that lesson to myself.

These non-publishable but essential scientific mistakes aren’t exchanged often enough in the lab, particularly before students graduate. Lab managers and technicians play invaluable roles here, but I think there’s more opportunity for sharing knowledge. There’s no single solution, and perhaps it even comes down to individual labs’ cultures.

But hey! This is the blog for digital scientists. How do you think we are going to tackle this problem?

Right, with software.

Software To Document Your Lab Know-how

Some web apps can help simplify tracking the problems resolved and know-how generated in your lab.

A simple solution is Evernote. There are many ways to use Evernote for science. One is to create a shared notebook and allow everybody in your lab to contribute new notes with details on how not to repeat their mistakes.

A more sophisticated alternative, Labguru, is a research and lab management service I helped build for managing experiments in academic biology labs. It provides one way to support and encourage a culture of learning as a team.

For each experiment, protocol, biological sample, reagent/kit, or any other research entity kept in Labguru, a comment thread retains notes that everyone has written. Lab members can then reference a particular item’s comment thread or search for relevant comments on similar items.

With this system, troubleshooting advice is easily accessible when a future problem arises or when a lab member wants to read about past experience before performing an experiment. This Labguru feature makes retaining a troubleshooting knowledge bank as easy as leaving a casual comment. It also helps celebrate results and methods that do work, which is of equal importance.

Would you spend 5 more minutes after successfully troubleshooting an experiment to share your new knowledge with your team? Would this peer-to-peer knowledge sharing actually keep students from repeating the same scientific mistakes?

Is this the way scientists should learn? 

 Making mistakes is required to do good science, but I argue that leaving others to repeat mistakes should not be part of the deal.

Change-to-win Scientific Mistakes

Making Change In The Lab

Changing a whole lab’s habits takes time, even if the objective is simply recording troubleshooting notes in a shared resource. Culture is a key to the speed and longevity of efforts to encourage team learning.

In a recent review of a science lab, Inc. Magazine offered a tip: “Be social. … The goal is to create a sense of connectedness so people want to help each other.” Indeed, scientists who hang out together learn together. Harvard Medical School researchers dubbed it the “water cooler effect”.

Simple changes can help encourage positive culture. Hang whiteboards in kitchen spaces. Write down a key thing you learned.

For example: “When you make your own Ligation Buffer do not skip ATP. Tried that. It didn’t work that well…” As silly as these notes might look, soon you’ll appreciate the fact that all the researchers around you do make mistakes. The question is how to make sure no one is stuck on an experiment for too long, that everyone is making progress.

A larger step can be to address scientific mistakes openly and compassionately as a group, perhaps during a regular lab meeting. This might be very hard to the person that is presenting the mistake made, but with proper management researchers can actually learn from it. Teachable moments present a great opportunity for introspection, to evaluate policies, and help encourage good habits in developing scientists.

Your lab will continue and evolve even after you leave. Your contribution can be extremely valuable to its future.

Resources Outside The Lab

Of course, sometimes the best advice comes from outside your lab and immediate network. Grad students in distress can also try posting questions to topical LinkedIn groups and even Twitter hash tags like #PCR.

Online forums offer more extensive help troubleshooting experiments. Here’s a shortlist of forums:

e3 — Epigenetics community troubleshooting, paper alerts, and commentary.

FigShare – Helps share unpublished results and facilitate dialog.

Molecular Biology Forum — Tips and questions for molecular and cell biology, microscopy, and bioinformatics hosted by the journal BioTechniques, which publishes a valuable exchange from the forum in every issue.

Protocol Online BioForum — Peer advice on a myriad of topics, from methods in biochemistry to zoology.

SEQanswers — For Sanger newbies to NGS experts, the forums (and associated wiki) are a great resource for troubleshooting sequencing experiments.

丁香园 and 生命科学论坛 — Two Chinese-language biology forums.


As these online forums demonstrate, we all desire to and need to learn together. But for some reason, our habits in the lab often impede our success.


It’s about time we change how we communicate our scientific mistakes.

[note color=”#dcdcda”]This is a guest post by Jonathan Gross (@rubp). Jonathan is the founder of Labguru, an online research management tool that helps you to plan experiments, track progress, and get results.[/note]


  1. Micropublishing platforms easily allow for the publication of FAILED reactions in my world of chemistry. But who’s proud of them? Who wants a DOI for one of those? Why not? They are not mistakes per se..they are “failures”. But most mistakes are failures too. Doesn’t research by definition guarantee some failures. Man up and publish them I say…okay….

  2. Chris Molloy, IDBS says:

    How refreshing – researchers encouraged to be open about their scientific mistakes while recognizing this information actually helps shape science! Jonathan Gross understands good science means making mistakes but that leaving others to repeat them is not part of the deal. Jonathan discusses tracking scientific mistakes for others to learn from, and looking at opportunities to share knowledge. All good stuff. Yet there’s so much more. It revolves around collaboration – a premise which beats at the very heart of science.

    R&D generates high value data assets; it’s the lifeblood of every laboratory. To avoid duplication and maximize scarce time and resources, quality data must flow through this knowledge ecosystem to thrive. Enterprise analytics require close collaboration to ensure metadata is captured with high context information. This contextualized data needs to be stored along with an ontology and its provenance. Experimental conclusions must be captured together with the intelligence of the community as they interpret and challenge, and social tools such as tagging, commenting and easy ‘sharing’ into the R&D data landscape can help create Jonathan’s desire for a sense of connectedness. Only if there is high context, connected stores can data be effectively aggregated and assimilated to generate a high quality information landscape.

    We may not know it yet, but we’re all in race, and it’s a race to achieve the highest context, highest value distributed datasets to make data reusable. Competitive advantage across all areas is much more about what data the analytics has to work on, rather than the choice of algorithm and filling in the data gaps. Who knows, with a more enlightened knowledge management environment, maybe that million dollar robot might have been broken only once instead of twice….!

    Chris Molloy, VP Corporate Development, IDBS


  1. […] Do you want to know what to do with not-so-exciting data and results? Check this post on how to document your scientific mistakes. […]