On Monday, July 1st, 2019 at 5:00 AM CT, I will write 100,000 words in a day and live stream the whole thing.
I plan to do this by dictating, as I explained in this blog post: Writing 100,000 Words in a Day.
In other words, I will speak 100,000 words and have the audio recording transcribed into written words. The majority of this transcription will serve as the first draft of my book, Sabbatical.
Given that I will be speaking somewhere between 12–16 hours, I am not speaking at all in the three days leading up to the event. Today is my second day of silence in preparation (more on that in a bit).
While most of my writing on July 1st will be for the first draft of my next book, I may also write on some other topics if it is necessary to achieve the 100,000-word goal (which I believe it will be). I don't imagine the book draft alone will need to be 100,000 words. My guess is that the book draft will be in the vicinity of 80,000 words. I have both some ancillary topics related to the book (which could turn into bonus materials) as well as some additional, unrelated topics ready, if necessary, to help me reach my word count goal.
When I'm delivering a focused, live presentation, I've found I usually speak at a rate of 10,000–12,000 words per hour. This is a greater intensity than what I would like to sustain for an extended period of time, however. My more relaxed, average speaking pace is between 9,000–10,000 words per hour.
If I budget a conservative average of 8,000 words per hour, I should be able to dictate 100,000 words in 12.5 total hours. I've decided to allocate a maximum of 30 minutes for each of the three meals (breakfast, lunch, and dinner). This additional 1.5 hours brings the total to 14 hours. If we factor in 6–8 bathroom breaks of 5 minutes each, and two 10-minute snack breaks, that will add another hour. This brings us to what I believe to be a conservative total run time of 15 hours.
Since I'm starting at 5:00 AM CT, that would take me to 8:00 PM.
Keep in mind, 15 hours would be as a result of going at what I would consider to be a leisurely pace of 8,000 words spoken per hour. Bringing this average up to just over 9,000 words per hour would shave off nearly 1.5 hours—which I think is quite doable.
Most of the 100,000 words I will write on July 1st are for my second book. The book is called Sabbatical.
I've been taking off every seventh week as a sabbatical since 2014, and it changed my life. Now, I want to change yours.
It is my mission, by 2047, to get every company in the world to pay their employees to take off every seventh week.
I want you to schedule your recurring sabbatical, and I'm going to teach you exactly how to do that in Sabbatical. This book will be your guide. You can learn more about the mission at sabbatical.blog.
Regular sabbaticals are the single greatest way to prevent burnout. There is no cure for burnout. There is only prevention, and sabbaticals are key to preventing burnout.
After speaking and writing on the topic of sabbaticals for the better part of five years, I know my message inside and out. It's only a matter of getting it out of my head and into book form. To that end, I needed to organize the thoughts in my head.
Over the past couple weeks, I've worked out an outline for the book. This list of bullets will serve as writing prompts for me on July 1st. When you see me looking at my iPad, that is what I'm looking at. The outline will let me know where I am in the process and remind me of the topics I want to cover.
Today is day 2 of 3 of no talking. I'm spending the three days leading up to July 1st in silence. I'm not speaking at all to rest my voice.
I’ve had a lot of calls and live streams this past week which have required me to speak for 4 to 5 hours per day. I not only need to give my voice rest to prepare for Monday but also to recover from the past week.
The main reason I'm resting my voice is out of preparation for July 1st. I want to go into Monday as fresh as possible to give myself the best chance of maintaining stamina.
I'm also prepared for the possibility of losing my voice completely, although I don't anticipate I will. Historically, I have done three long, consecutive days of recording for the Overlap audiobook. I maintained a pretty conservative pace, and topped out at around 6 or 7 hours recorded in one day. I was definitely feeling my voice by that point, but I hadn't lost it. I could have kept going, but we already had three days set aside for the project, so there was no need to be a hero. Consider also that with the audiobook, it was important for the finished result to sound good. So I was definitely projecting my voice more than I will need to here.
If I do lose my voice on Monday, I will switch to typing and type as much of the remainder as I can.
I used to get so caught up in my work that I would forget to drink water. So I started tracking my water intake.
In the past, I used a Bluetooth water bottle. Now, I use an app called WaterMinder. I like it a lot. I actually drink more water because of it, and that was more than worth the $5 I paid for the app.
I'm focusing intently on staying hydrated in the days leading up to the Write100K event. I plan to keep my coffee consumption to a minimum, since it is a diuretic and will result in increased frequency of bathroom breaks—which will steal time and dehydrate me. Caffeine also dries out the vocal cords.
I do drink coffee every day, but I've found that I am usually lucky enough not to get headaches on the rare days I skip coffee. I plan to overcompensate in the area of sleep to address my energy needs (more on that in the next section).
On the off chance I do get a headache from going without caffeine, or if I just need a pick-me-up, I'm okay with allowing myself a glass or two of cold brew coffee. I just don't want to overdo it.
Regular listeners of the seanwes podcast know from The Sleep Episode that I am serious when it comes to resting. I sleep 8 hours every night.
My event starts at 5:00 AM CT, so I plan to wake at 3:00 AM, run a few miles, stretch, and shower. This should leave me with about a half hour to prepare mentally. I like to sit in my bean bag and meditate.
While I normally sleep 8 hours per night, I want to get even more sleep the night before my Write100K event. I'm shooting for 9 hours.
- If I’m to wake at 3:00 AM on Monday, and get 9 hours of sleep, I need to go to bed at 6:00 PM.
- To be tired enough to fall asleep at 6:00 PM, I need to wake at 3:00 AM on Sunday.
- To wake at 3:00 AM on Sunday and still get my usual 8 hours of sleep, I need to go to bed at 7:00 PM (that’s tonight).
Last night, I went to bed at 8:30 PM and got up at 4:30 AM this morning. I was originally shooting for 8:00 PM–4:00 AM, but since it ended up being 30 minutes later than I intended, I went for a run between 5 and 6 miles to help me fall asleep more easily.
I've taken time each day in the past week to visualize myself performing this feat.
This is not unlike an athletic event. I'm certainly treating it like one.
I spent a large part of my run this morning visualizing every aspect of what will happen on Monday:
- Waking up at 3:00 AM and feeling rested from 9 hours of sleep.
- Going for a run (and the specific route I'll take, which will be lit by street lamps).
- Stretching and foam rolling.
- Entering my office at 4:30 AM with the black night sky still visible through the slats of the blinds.
- Turning on the cameras and lights.
- Sitting in my bean bag and focusing my thoughts.
- Watching the 60-second countdown timer on the live stream at 4:59 AM.
- Seeing the camera shot on the video monitor switch to me in my peripheral vision as the event starts.
- Speaking at a low, natural volume (not over projecting).
- Taking a break at 6:00 AM to eat breakfast (which will be a bagel and eggs).
- Looking at my outline on my iPad mounted in the stand.
- Saying, "Hey Ben, bathroom break," and going to the bathroom after the "Bathroom Break" slide appears on the screen.
- Sitting in the bean bag and swinging the microphone on the stand in front of me, with an iPad in my lap.
- Taking a break at 11:30 AM or 12:00 PM, depending on my hunger level, to eat lunch (which will be a turkey sandwich and apple).
- Continuing to speak into the afternoon and starting to feel my voice fatigue but remaining focused.
- Drinking room-temperature water.
- Taking snack breaks (which will be nuts and dried fruit).
- Taking a break at 5:00 PM for dinner (which will be spaghetti and french bread).
- Seeing the sun start to dip in the sky through the blinds.
Boy, is there a LOT of technical stuff that has to happen in order to give you the live show you're going to get on Monday. I've been planning for months.
Believe it or not, I was planning to handle everything related to live streaming the event myself. This includes camera switching, audio recording, live word count, and more.
I'm no stranger to live streaming (I single-handedly stream the video recording of the seanwes podcast every week and handle recording and switching of cameras myself).
But all of this would have been on top of my having to also write 100,000 words. That's just a bit too much.
Fortunately, I wised up and decided last week to hire my podcast co-host, Ben Toalson, to spend the day with me and handle all things technical.
We had a meeting last Monday going over all of the technical details that lasted over 3 hours.
The investment will be worth it. Ben's goal is to make it so I can just walk in at 5:00 AM and start—and not have to deal with any of the technical stuff. If that ends up going as planned, it will be a huge mental weight lifted off of me. The technical stuff related to live streaming and recording a multi-camera show just for the normal podcast each week is already mentally taxing enough—without the added complexities of the Write100K event.
Part of me is still worried that I'll need to solve a technical problem during the event that Ben doesn't know how to fix, but I'm trying not to think about it.
Live Word Count
I'm really proud of the solution I came up with for this. It's actually a huge technical feat.
Let me first describe what's needed.
I want a way to show a total word count on the live stream that is at least fairly up-to-date. I had a number of ideas for how to achieve this, but none were ideal, as they all interrupted the flow of the recording, would require me to stop talking, could take 5 to 10 minutes to process, and would only really be possible during breaks. This was unacceptable.
I needed a way to get word count updates whenever we needed to, without interrupting my delivery, and without having to wait for audio files to export, and I finally figured it out how to do it.
We are recording my local audio in Logic Pro X. This has the master project we'll use to create the final audio file to deliver to the transcriptionist.
I'll have two different microphones for various positions I'll be in during the recording, and I did some aggregate device magic to make sure the audio from both input sources is captured on the same track. This is critical for the next step I'm about to describe.
Since all input sources will be captured on the same audio track, all Ben has to do during the live stream is stop and start the local recording in Logic Pro X in between two of my sentences. This will create a new .aif audio file in the project package. He'll have a Finder window open with the package contents so he can view the individual source audio files.
While the new audio file is recording, Ben can drag and drop the previous audio file into Descript (my favorite transcription service).
I waxed eloquent about Descript in my last blog post, so I'll quickly quote you what I said before:
Descript is an app that automatically transcribes any audio file in almost no time. In minutes, you have a full transcription. But this is no ordinary computer-made transcription (more on that in a minute). No longer do I have to wait days for a transcription service to return a text file. Instead of the industry standard $1/min, it's only $0.07/min.
If you want to try it out, here's a link to get 100 free minutes with Descript.
Descript is simply magic. I know they use the Google Speech engine on the back end, but some kind of voodoo happens where the computer-made transcript actually has punctuation. Yes, we're talking intelligent commas, periods, etc. All ready within minutes. It's not like Dragon Dictate where you have to say "Period. Comma. New Paragraph." Descript figures all of that out intelligently.
- Speak naturally.
- Drag and drop your audio file.
- Get a fully punctuated transcript.
I use it every day, and I love it.
I record most of my conversations because I can always drop the recording into Descript to get a full transcript of what I said. From this transcript, I can create all kinds of other content: emails, blog posts, videos, social media posts, graphics, speeches—you name it.
What's remarkable about Descript is the accuracy of the transcription. Maybe it's because I use a quality microphone, but the transcription accuracy is darn near 98%, I'd say.
I repeat: in minutes you have a full, punctuated transcript that is 98% accurate.
All Ben has to do is select all of the transcribed text in Descript to get a total word count. He can then add that word count to the existing one in our live streaming software to update you with a fresh word count any time he wants.
Descript is the secret sauce that makes this possible.
I use Descript to create transcripts and captions for all of my podcasts and social media videos. At 98% accuracy (this is my estimation, by the way), I usually only need to clean up a few things and it's good to go.
However, 98% accurate isn't going to cut it for my actual book draft. That's why I plan to use a human transcription service to transcribe my writing from the Write100K event. At an average of $1/minute, it's likely the transcription will cost me nearly $1,000, but it's worth it.
I'll use Descript mainly just for presenting you with the live word count on July 1st.
Oh—and beyond just providing you, the live-stream viewer, with that courtesy, this live word count will be equally helpful to me for knowing how far along I am in progressing toward my goal!
I'll have four cameras set up for the Write100K event to allow me to switch positions throughout the day.
- Camera 1: Sitting shot.
- This is the normal shot I use for the seanwes podcast.
- Camera 2: Standing shot.
- This is for standing in the same place as my sitting shot. Since I'll be in the same place, I can use the same microphone and just raise it up on the boom arm.
- Camera 3: Bean bag shot.
- This is for when I want to lounge in a more comfortable position.
- I will use a second microphone on a mic stand for this shot.
- I will have a foot switch for toggling the mute on this microphone so as not to introduce unwanted noise or feedback when not in use.
- Camera 4: Room cam.
- This is a shot of the entire room, courtesy of the GoPro in the corner near the ceiling, for establishing context and adding visual interest.
I'm not recording on any of the cameras themselves. It would take up far too much storage and is simply not necessary.
Instead, I am recording the live feed in 1080p locally onto a 32TB network storage device where there is plenty of room. The cameras run into the Mac Pro via HDMI to SDI to Thunderbolt.
Audio is the main thing I’m concerned about, and that’s being recorded in Logic Pro X. I’ll also have a backup hardware audio recording, and then there’s the YouTube live stream which serves as yet another fallback. So that’s four places audio is being recorded.
I want to also give a shout out to my wife, Laci. She will be taking care of food and drinks throughout the day, for both me and Ben, to make sure we stay nourished.
Dan, on the seanwes team, will also be helping tend to the live chat throughout the day, as I will obviously not be able to pay any attention to it. He'll answer any of your questions when he can.
I'm writing this post mainly for myself to capture this moment in time, and to remember what I did and what went into the Write100K project.
Overall, I'm quite optimistic and excited. This is something I've been planning to do since the beginning of 2019. It was originally going to happen in May, and then we had to reschedule it for July, so I'm glad to finally do this.
I am motivated by the support and encouragement I've received. I really do feel like I can do this, but part of me can't help but feel surprised at everyone else's confidence in me.
Do they really just not know how many 100,000 words actually is? Or have they really processed it and actually believe that much in me?
The main thing I will have to look out for, and make a conscious effort to avoid, is overly projecting my voice. After recording somewhere close to 1,000 podcast episodes, I've developed a certain "broadcast voice" wherein I project more than I would if I were just having a casual in-person conversation with a friend. I shape my voice to sound good and create an enjoyable listening experience—much like a singer projects when performing in front of a crowd. This is fine for a 1–2 hour show, but isn't sustainable over an extremely long period of time.
I have to continue reminding myself that I only need to optimize for the transcription. It doesn't need to sound good, I don't need to project, I'm not trying to create an optimal listening experience; I need only speak with enough energy so as to enable accurate transcription. Minimum Viable Clarity.
I'm going to forego using any monitoring headphones (which I normally wear while recording so as to shape my voice and ensure a good sound) in order to focus on maintaining a calm and relaxed tone that I can maintain for an extended period of time. This will afford my voice the best chance of lasting long enough to accomplish my goal of dictating all 100,000 words without having to resort to any typing.