Read more of this story at Slashdot.
What if the way your team is connected matters as much as the people on it? Damon Centola, author of Goodthink, joins Vasco Duarte to explain why networks do more than move information: they shape trust, legitimacy, coordination, and creativity. This conversation gives software leaders a practical way to think about AI adoption, Slack, and the conditions for better collective thinking.
"The networks aren't actually pipes, they're prisms."
When leaders announce a new tool or practice, the message does not land in a vacuum. People interpret it through colleagues they trust, through the risks they see in changing their work, and through the signals that tell them whether the change is legitimate. Centola explains why a campaign that simply broadcasts information can fail: behavior change needs social reinforcement, not just a clearer message.
"It's not just about teaching people facts, it's about getting them to believe and want to believe."
For software leaders introducing AI or another new way of working, the useful question is not only how to communicate the change. It is whether people can see peers like them trying it, adapting it to their work, and making it safe to join. Broad bridges between groups help a practice cross from one team into another because people can build trust and confidence together rather than being sold an idea by a broker.
"The kinds of structural solutions that facilitate and answer the coordination challenge would actually undermine the creativity challenge."
Fast, highly connected channels can be exactly right when a team needs to coordinate under pressure. They can be harmful when a team needs to explore, because everyone quickly converges on the first promising answer. Centola suggests treating organization design as something fluid: amplify fast coordination when the work calls for it, then create more space and separation when the next challenge is discovery.
"By slowing down the learning process in terms of peer learning, you accelerate the local discovery process."
Centola's experiment with data scientists compared fully connected teams with groups organized more like hallways. In the fully connected teams, everyone saw the current best answer and rushed toward it. The hallway networks allowed different approaches to develop for longer; every one of those networks outperformed the fully connected networks across eight competitions. The lesson is not to abandon collaboration, but to protect enough diversity for people to explore beyond the obvious local peak.
"We really are these geographic beings. We live in the spaces that are immediately around us."
MIT's Building 20 became an unlikely source of creativity because people from unrelated disciplines kept running into one another without being pulled constantly back into their home departments. The same principle applies to digital workplaces. Leaders can ask which connections help people coordinate, which ones create conformity, and where a smaller set of unexpected conversations might give an idea time to grow.
"All the studies I've ever done are up there, but also all the data, all the scripts, so you can kind of dig in."
At the Network Dynamics Group, Centola and his colleagues connect fundamental network research with practical work in organizations, hospitals, and civic life. Their work includes network-based clinical consultation designed to reduce bias and medical errors. For listeners who want to go deeper, the group shares papers, data, and scripts alongside the research.
[The Scrum Master Toolbox Podcast Community CTA]
Damon Centola is the Elihu Katz Professor of Communication, Sociology, and Engineering at the University of Pennsylvania, director of the Network Dynamics Group, and author of Goodthink. His research studies social networks, behavior change, and collective intelligence, including how structure shapes decisions. LinkedIn: [missing]. Check out Damon's work at the Networks dynamics group of the University of Pennsylvania.
You can connect with Damon through the Network Dynamics Group. No LinkedIn profile or Twitter/X account was provided in the interview material.
I use Camtasia to record my courses when I partner with Pluralsight or LinkedIn Learning.
Every recording ends up with long pauses while I think, lip smacks and mouth clicks the microphone picks up, or a mouse click or keyboard clatter right in the middle of a quiet gap.
Cleaning that up by hand is dull.
Scrub the timeline, zoom in, ripple delete, repeat. Multiply by a 60 minute course and it becomes a big part of the job.
A typical 60 minute course takes me 30 to 40 hours to create from start to finish. This initial clean-up pass alone used to eat about 6 of those hours. I decided to build a small Windows app to do this initial cleanup phase for me.
You point it at a Camtasia project, it finds the noise, you review what it found, and it writes a new project with the edits already applied.
~
Camtasia Cleanup takes a .tscproj file and removes the stuff you’d otherwise cut out by hand:
Full transparency – it’s not a magic “make me sound professional” button. It’s a time saver for the dull first pass, and you stay in charge of what gets removed.
One rule I stuck to from day one: it never touches your original project or raw recording. Everything is written next to the input with a -cleaned suffix, so the worst case is you delete the new files and carry on as before.
~
There’s no AI model or cloud service involved. It’s classic audio analysis, which means it’s fast, it’s repeatable and nothing leaves your machine.
.tscproj file is JSON. The app reads the clips, the recording each one points at, and anything layered on top (b-roll, callouts).The slow bit, the analysis, runs once and is cached. After that, moving a slider or unticking an edit re-plans instantly.
~
The main features:
Now, how to use it.
~
1. Open a project. Click Open project… or drag a .tscproj file anywhere onto the window. If the recording has moved, the app offers to let you locate it.
2. Analyse. Hit the green button. It runs once and is cached for next time. Here’s a short 2 minute clip from one of my courses once the analysis has finished:

The waveform shows what it found. Red marks a cut pause, orange is a click that will be removed, blue is a place it kept because the screen was changing. The list on the right has every edit, and the bar at the bottom tells you what you get: 2:03 down to 2:01 here, 4 pauses shortened, 28 clicks removed.
3. Tune it. Move the sliders or untick rows. Everything re-plans instantly from the cache, so you can experiment freely. I find I don’t need to use that so much.
4. Check the result. Switch to Cleaned to see and hear the result. The playhead stays on the same moment, so you can jump straight to something that sounded odd.

5. Export. Click Export. Next to your original you’ll get:
yourproject-cleaned.tscproj, which you open in Camtasia as normalyourproject-cleaned-audio.wav, the cleaned narration (one per clip for multi-clip projects; a clip that also has a played file gets -narration and -system files on two tracks)Open the -cleaned project in Camtasia and carry on editing.
~
After the first version I still did a manual pass in Camtasia on every clip. So I gave Claude Code two project files: the raw recording and the cleaned version I had edited by hand. The question was simple: what do I keep fixing that the app misses?
The answer was in the gaps between words.
On a slide-deck clip I had muted every gap of 0.7 seconds or longer that the app left in: 26 out of 26. Almost all of them were the pauses it keeps because the screen is changing. That became the optional “silence long gaps” rule. It also showed that four of the five inhales the app flagged were ones I had muted by hand, which gave me some confidence in that rule.
I held back on turning it on by default, because of something my own recordings made obvious. A slide-deck pause is usually deliberate. An IDE recording has long gaps while I type, copy and paste, so silencing them blindly is a different decision. So it is off by default, it skips any gap that is not really quiet, every gap is a row I can untick, and I checked what it would do on an IDE clip before trusting it.
~
There are two savings here, and they’re easy to mix up.
1. Less recording to deal with. The app cuts the dead air out for you. On one of my course modules it took the recording from about 10 minutes to under 8, before I’d touched a razor blade. That’s more than two minutes of pauses and noise I never had to find, zoom into and delete. Even the short 2 minute clip in the screenshot above came down from 2:03 to 2:01, with 4 pauses shortened and 28 clicks removed. Each of those is a cut I’d otherwise have made by hand.
2. Less of my time. A typical 60 minute course takes me 30 to 40 hours from start to finish: planning, scripting, recording, editing, reviewing. The app saves me about 6 hours of that editing. That’s roughly 15 to 20 percent of the whole job, from one tool that handles the dullest part of it.
The workflow is: open the project, analyse (once, then it’s cached), skim the list of edits, untick anything I disagree with, export.
Six hours is nearly a full working day back on every course, time best used on other tasks.
The main time saving is in the dull first pass and I still listen to every cleaned clip. So that is still a thing. It’s a great first pass but it’s not the finished edit. The 6 hours is my own estimate across my courses, not a stopwatch measurement.
~
A few caveats:
~
It does what I need it to do.
Open your project, analyse, review, export, carry on editing in Camtasia and go!
If you record in Camtasia and you’re tired of the same clean-up routine, it might save you some time too.
~
Enjoy what you’ve read, have questions about this content, or would like to see another topic covered?
You can schedule a call using my Calendly link to discuss consulting and development services.
~
Check my AI courses. From developers to decision makers, these have you covered:
~
Welcome to F# Weekly,
A roundup of F# content from this past week:
Microsoft News
Videos
Blogs
Highlighted projects
New Releases
That’s all for now. Have a great week.
