I thought this would turn out to be the toughest part (see part 1 - the website ). Four people, thousands of miles apart, recording over Skype. With a host who hasn't hosted anything since I recorded 'radio shows' on a tape deck aged about 9. But it went very smoothly - The credit for this goes to a) my guests and b) Skype.

Guests

I had about 40 people register as being interested in guesting. I didn't have a great deal of time to research people, so I picked people from their profiles to suit the top stories I thought we'd discuss. I also wanted a mix of people. Broadly I am going to aim for the following:

  • Location: 1 SF, 1 Other US, 1 Non-US. As someone pointed out in a HN comment, it's easier to follow a Podcast with multiple accents. I loved Matt Brace's Texas accent in our first show, and Luzius Meisser's Swiss in our second.
  • Background: 1 YC, two others - not entirely sure yet - I don't want it to be all YC, and I'm averse to people already knowing each other before guesting.

I was terrified that I was exceptionally lucky with my first guests. I edited nothing out - every word they said is untouched and uncropped - all three were outstanding speakers. But the second show also worked really well, so hopefully the formula is there.

Content

I don't know if I have the content quite right yet. Two major topics seems to be about right in terms of length. It's hard to know which stories to go with. In the first show we talked about Kickstarter, the Raspberry Pi and briefly about Instagram. We also discussed HTML5 vs Native apps. I cut this out because I wanted to keep to a 30 minute podcast - on reflection that might be an arbitrary number. I'm open to suggestions here: Should I just release however long is usable? Stick to 30 minutes religiously? Or release a 30 minute one, and then a 'bonus extra' of the rest?

I emailed everyone the day before with the topics.

Scheduling

Timezones are an issue. I googled 'timezone group scheduling' and found Doodle, which worked extremely well. (Although I did have to manually add 48 separate timeslots) 5pm in the UK is 9am in SF. London and San Francisco were my two geographical extremes - if I go further East than Europe then I am going to be in trouble. I had a 48 hour window to find a mutual timeslot - and in the end we had a two options. We recorded at 6pm UK/10am PDT.

Recording

We held a group Skype. Everyone recorded their own end of the conversation using whatever tool was to hand (Audacity or Garageband). Some people used Audio Hijacker to isolate down to just their own microphone, but I seemed to be able to just hit record in Garageband. I was in fear that my computer would freeze and I'd lose the conversation and the whole recording. In future I might divide the chat into two to mitigate this risk.

I also recorded the group call as a backup in case someone's recording didn't work out. Then everyone Dropboxed (Tangent - using a company name as a verb = >$1bn valuation) their files to me and I lined them all up. At the start we said 1..2..3.. and then clapped so that I could sync things later on in editing.

Skype introduced quite a lag - perhaps one second. Having separate tracks let me remove this lag by realigning them.

Something that I hadn't appreciated, and that I need to work on, was the delay that people left by waiting to see if anyone else spoke - sometimes five seconds of silence. I need to ask people by name more, maybe editing this out later.

I tried Garageband for editing but couldn't figure out the UI at all, so I went to Audacity and it worked brilliantly.

Once I'd realised I could sync-lock tracks together there was no stopping me. I sliced and diced my own track until I sounded coherent, chopped out most of the silences. Rearranged answers to make them more coherent. I silenced closing doors and squeaking chairs, tightened up pauses, and removed a lot of mmmms (virtually all my own). It felt amazing being able to change what people say - far more magic than photoshopping. It is easy to remove single words from even quickly spoken sentences. Editing took about two hours.

Then I threw in a bit of a jingle at the start.

Hardware:

A Macbook Air, Some headphones and a £40 microphone from Maplin (A UK equivalent of Radio Shack). I'm not sure anything more is necessary. I think the other guys all had good quality microphones too, probably far better than my own.

Lessons:

I sat on the idea for almost a year. For no reason other than inertia. There's no magic to it. Just get off your backside and start doing something - I'm delighted for a first attempt and hopefully it'll just continue to improve.