Jan 21

Venture Unlocked #3

I’ve decided to spend some time each week in Q1 trying to work on the craft of venture capital. While I’ve been doing this for close to a decade, I’ think there is always room to learn more. To that end, I’m going to be listening to Venture Unlocked by Samir Kaji, a fellow Kauffman Fellow who is looking at venture from a unique angle given his seat at First Republic.

Since I’ve already listened to the first couple episodes I decide to start with episode 3, Roger Ehrenberg.

Given the notes provided in the link above, I won’t bother telling you what’s in the episode, instead I’d like to give my analysis.

  • To find his partner Roger took 1000 applications, did first calls with 100 people, interviewed 20 and whittled it down to 5 finalists before choosing one person. I’m a fan of such a rigorous process.
  • They actually found fund size growth painful. The fund sizes went 50M, 105M, 160M, 160M. That’s a pretty quick progression up to a steady state.
  • They are pretty aggressive about recycling but that has gotten easier since they invest cross fund. A series B in a later fund get’s liquidity earlier than a seed and makes recycling in that fund easier. I really like that observation a lot since cross fund investing is a controversial idea.
  • When it comes to flexing to get into a deal, they keep capital at risk steady and instead take a lower ownership with an eye towards buying more later. I’m a fan of being disciplined about check size, especially for smaller funds.
  • Their biggest learning was essentially the power law rule: Great companies can get much larger than you could have ever imagined out the outset. Even though everyone believes this, it feels like you have to experience it for yourself to really grok it.
  • Once they had the returns to be picky they were very deliberate about their LP makeup. It’s not often you hear about funds breaking up with LPs, so this is some good insight into what they looked for. Not surprising mind you, funds that are in it for the long haul.

Overall, this is a really good conversation about the growing pains of taking a fund from proof of concept to steady state. My big takeaways were that they learned a lot on the fly and try to do small experiments to keep improving over time. Seems like a team with a real growth mindset.

Dec 20

My Top Beers of 2020

For the last couple of years I’ve logged almost every beer I’ve had on Untapped. It’s both been a fun way for me to learn more about beer and my own palette. For example, I’ve learned that I particularly like Mosaic hops.

The data also allows me to know what my favorite beers of the past year were. And I thought I’d share the top five for kicks.

Untapped ranks beers with the same score by the number of times you had them, so it probably shouldn’t be surprising that local breweries that I can easily get are over represented. As an example, I actually like Smooth Beats Miami more than DDH Oh, but it’s not as easy to get, so it came in third rather than second.

Anyway, this was just a fun little post to get me started blogging again. See you in the new year.

Jun 20

Cyber Thoughts May 2020

Well, I’ve forgotten to put the last few of these up here. Instead of putting the content here I’ll just put a link to the newsletter and let you signup below if you wish.

Cyber Thoughts May 2020

Sign up here:

Mar 19

Cyber Thoughts March

Here is the latest Cyber Thoughts Newsletter.

View in new tab

Feel free to signup if you’d like to receive them directly:


Feb 19

Cyber Thoughts February

Happy V day. Here is the Cyber Thoughts Newsletter I wrote for February.

View in new tab

Feel free to signup if you’d like to receive them directly:

Jan 19

January Cyber Thoughts

As part of communicating with investors and friends of our new fund, I’m writing periodic newsletters about what we find interesting in cybersecurity. Below is the one I wrote for January.

View in new tab

Feel free to signup if you’d like to receive them directly:

Jan 19

Lytical Ventures

Wow, time flies.

About a year ago I launched a new fund with a great group of partners. I’m not going to write to much about it here at this time but it’s an early stage venture initiative, targeting the cyber and corporate intelligence landscape. You can look up more about it on LinkedIn.

Suffice to say that I’m really excited about it. Hopefully, now that we are well on our way I’ll have more time to write posts here.



Dec 17

NYC Mesh

Given that it looks like Net Neutrality is going to be killed shortly I decided to look into find a non-censored Internet Connection. NYC Mesh is a pretty interesting volunteer project doing just that.

I signed up and one of their team came out and did a site survey. It looks like I’m a bit too far from the Brooklyn Supernode for a direct connection, but hopefully we can find a 1 hop solution in the coming weeks. If you’re interested in having a net connection not controlled by a media conglomerate I suggest you check them out and get involved.

Dec 17

I Had a Kid

This really isn’t a personal blog, but this is pretty life changing so here goes.
Milo Young Nelson was born around 12 weeks ago.

I’m finally getting back to a somewhat normal sleep schedule and I’m digging into some fun ideas. More soon.

Jun 17

Golem – The Blockchain Supercomputer

tpsdave / Pixabay

Today’s rabbit hole for me is Golem, a group trying to monetize unused processor cycles. Again, this is a throwback to the ideas of, among others, MojoNation. Their alpha will allow Blender users to render CGI in a distributed manner.

The problem with distributing computation across a high latency network like the Internet is that you need problems to be extremely parallel, since almost any amount of communication between nodes is uneconomical. This is fine for things like SETI@home but isn’t great modeling that requires the previous output as input. Two major questions I have: are there enough use cases where someone has the ability to pay for this type of computation where they aren’t worried about have someone see the work they are doing and will the security model be strong enough to get people to run unknown code on their machines.