Author Archives: kevin

About kevin

I write the posts

Let employees sell their equity

Posted on

Sometimes people choose to work for one company over another for reasons related to the work environment, for example what the company does, and whether the other employees create a place that's pleasant to work at. But a major factor is compensation. If Company A and Company B are largely comparable, but Company A offers $30,000 more in base pay per year more than Company B, most people will choose Company A.

At tech companies, compensation usually breaks down into four components: company stock, benefits, cash salary, and bonus. When you get an offer from a company, these are the four areas that the recruiter will walk you through. The equity component is a key part of the compensation at startups. Small startups hope that the potential for a large payoff is worth sacrificing a few years of smaller base pay.

If you join a small startup and you get stock, you generally can't sell it until an "exit event" - an IPO or acquisition - even if your entire stock grant has vested. Generally, any stock sale before an exit event will require approval of the board, and the boards generally frown on stock sales, for reasons I will get into. So while you may own something that is worth a lot of money, you can't convert it into cash you can actually spend for a half decade or more.

By contrast, if you join a public company, your compensation includes equity that you can sell basically immediately after it vests, because it trades on a public exchange. There are hundreds of people who will compete to offer the best price for your shares every day between 9am and 4:30pm.

As an employee, how should you think about the equity component of your offer? One reason to take a big equity stake is to bet on yourself. If you have a great idea about how you can make the company 10%, 50%, or 200% more valuable, and you think you can execute it, you should take an equity stake! After you implement the changes, your equity will be massively more valuable. Broadly speaking this is what "activist investors" try to do; they have a theory about how to improve companies, they buy a stake and hope the value changes in line with the theory.

One problem with this is that you are much likely to be in a position to make these changes if you are someone important like a C-level executive or a distinguished engineer. However, most tech employees are not C-level executives. If you are an engineer on the fraud team, and you try really, really hard at your job for a year, maybe you can increase the value of the company by 1% or 2%. You are just not in a position, scope wise, to drastically alter the trajectory of the company by yourself.

Rationally speaking, it does not make much sense for you, an engineer on the fraud team, to double or triple your effort just to make your equity stake worth 1% more. There might be other reasons to do it - you could really buy into the mission, or you hate being yelled at or whatever - but just looking at the compensation, whether you, personally, work really hard or slack off, your stock is probably going to be worth about the same. Unless you are the CEO or other C-level executive, at which point you have a big enough lever that your level of effort matters.

Another way to think about it is, imagine you have invested your money in a broad range of stocks and bonds, and then someone asked you to sell 30% of it and place it all in a single tech stock. Modern portfolio theory would suggest that that is a bad thing to do. You could gain a lot if the stock does well, but on the other hand, if the company's accountant was embezzling funds, or the company lost a lawsuit, or the company lost a database or had the factory struck by lightning or something, you could lose a ton of money that you wouldn't if you were better diversified. It's not worth the risk.

All this goes to say that employees should value their equity substantially less than an equivalent amount of cash. Outside of the C-level, you can't do much to make the equity more valuable, and an extra dollar worth of equity takes your portfolio further away from an ideal portfolio that you could buy if you just had cash. (For more on this topic you should read Lisa Meulbroek (hi, Professor Meulbroek), whose CV is criminally underrated.)

(On the flip side, if your company is small and valuable, it may have its pick of investors to take money from, and be able to dictate investment terms. Holding equity in a company like this is a way to approximate the "deal flow" of a good Silicon Valley investor - as an employee you are getting the chance to buy and hold stock in a company at prices that would not be accessible to you otherwise. This may be true of small, hot startups but it gets less and less true the bigger a company gets and the more fundraising rounds it goes through.)

One implication is that you should prefer to work at public companies. At a public company, you can take your equity compensation and immediately sell it and buy VT (or even QQQ) or whatever and be much better off because you are diversified. You can't do that at a private startup.

Another problem is that public companies tend to have better equity packages. I went through a round of interviews recently and I was stunned at how paltry the equity offers were from private, Series A-C companies. For most of the offers I received, the company valuation would need to increase by 8-20x for the yearly compensation to achieve parity with the first-year offer from a public SF-based company, let alone to exceed it. Even if they did achieve 4 doublings of their valuation, you might not be able to sell the private company stock, so you're still behind the public company.

I expect larger companies to have better compensation, it's part of the deal, but that large of a differential, plus the cash premium to be able to sell instantly, makes it foolish to turn down the public company offer. 1

So how can you compete if you're a smaller company? The obvious answers are what they've always been: recruit people with backgrounds that bigger companies overlook, give people wild amounts of responsibility, sell people on the vision, commit to "not being evil" and actually follow through on it.

But you can also try to eliminate an advantage that public companies have by letting your employees sell their equity. Not just, like, one time, at a huge discount before you go public, or when you get to Stripe's size and want to appease your employees. But routinely; because your employees want to boost their cash base, or buy the stock market, or buy a vacation, or whatever.

There are some objections. Having more than 500 shareholders triggers SEC disclosure requirements, which can be a pain to deal with. So require employees to sell to other employees or existing investors. Cashing out entirely might send the wrong signals, so limit sales to 10-20% of your stake per calendar year. A liquid market might require repricing stock options constantly. So implement quarterly trading windows.

Executives might not want to see what the market value of your stock is at a given time. That's tougher. But a high day-to-day price might convince people to join when they otherwise wouldn't. A low price might convince you to change direction faster than waiting for the next fundraising round.

There are also huge benefits. Employees can cash in earlier in ways that are generally only available to executives. They can take some risk off the table. People who want to double up on their equity position can do so.

Finally, you might be able to attract employees you might not otherwise be able to. A lot of folks who are turned off by the illiquidity of an equity offer might turn their heads when you describe how they can sell a portion at market value every year.

Big companies have big moats. One of them - the ability to convert stock to cash instantly - doesn't need to be one.

Thanks to Dan Luu and Alan Shreve for reading drafts of this post.

You may think they were lowballing me, but this was after negotiation with each. Another possibility is that I did differently on the interviews for each, and the smaller companies offered me lower packages because they thought I did worse. I think I did about equally well on the interviews for each.

Liked what you read? I am available for hire.

Building a better home network

Posted on

I finally got my home network in a place where I am happy with it. I wanted to share my setup and what I learned about it. There has never been a better time to set up a great home network; there are several new tools that have made this easier and better than in the past. Hopefully this will help you set up your home network!

My house

My house is two stories on a standard 25 x 100 square foot San Francisco lot. The ground floor looks roughly like this:

 --------------------------------------
|               |                      |
|               |         |   Office   |
|    Garage     | Mudroom |            |
|               |         |-------------
|                           | | | | | |
 ---------------------------------------

Upstairs looks like this:

 --------------------------------------
|    ___________                       |
|               |        Living Room   |
|    Bedroom    | Kitchen              |
|               |         -------------
|               |           | | | | | |
 ---------------------------------------

We have a Roku in the living room. My goals for home internet were:

  • Good wireless connection in every room
  • Ethernet connections in the office
  • Ethernet connection to the Roku
  • Synology network attached storage (NAS) and other external hard drives reachable from anywhere in the house

We are lucky to have Sonic Fiber internet service. Sonic comes in to a box in the garage, and an Ethernet line runs from there to the mudroom. None of the other rooms have Ethernet connections.

Initial setup

Sonic really wants to push Eero routers to everyone.1 Eero is fairly easy to set up, and Sonic collects a small fee from renting the router to you. You can extend your home network by adding more Eero's into a mesh network.

If you have a small apartment, an Eero is probably going to be a good fit. However, the mesh network was not great for achieving any of the goals I had in mind. The repeaters (Eero beacon) do not have any Ethernet out ports. It was difficult to extend the network from the mudroom to the bedroom without renting two extenders, which added about $100 per year, increased latency and lowered speeds. Further, clients on the network kept connecting to an Eero that was further away, instead of the closest one.

Powerline

(NB: please don't stop reading here as I don't recommend this.) My next step was to replace the Eero's with a traditional Netgear wireless router in the mudroom. This also could not reach to the bedroom. So I bought a powerline adapter and plugged one end in near the router and the other end in the bedroom.

Powerline adapters send signal via electric current in your house. They don't offer great speeds. Devices on your network that use a lot of electricity, like laundry machines or the microwave, can render the powerline connection unusable.

There are probably better solutions for you than powerline adapters in 2020.

Extending Ethernet to more rooms

I called a cabling company about the possibility of running Ethernet to more rooms in the house. We decided the bedroom would be very easy since it's directly above the garage. It took a team of two two hours to drill a hole in the garage, run a cable up the side of the house to the bedroom, and install an Ethernet port in the bedroom. This cost about $200.

We looked at running Ethernet to other rooms but the geography of the stairs made this really tricky.

Side note: future proofing cabling

Our house has coax cables - the traditional method of getting e.g. cable TV service - running from the garage to four rooms in the house, but it doesn't have Ethernet set up. This is disappointing since it was built within the last decade.

There are two things you can do to future proof cable runs in your house, and ensure that cables can be replaced/swapped out if mice eat them or whatever. I highly recommend you implement them any time you are running cable. One is to leave a pull cable in the wall next to whatever cable you are installing. If you need to run a new cable, you can attach it to the pull cable, and then pull it all the way through from one end to the other.

Normally cables will be stapled to the wall interior, which makes them impossible to pull through. The other option is to leave cables unstapled. This will let you use the coax/other cable directly as the pull cable. In general though it's better to just leave a second pull line in the wall behind the port.

Without either of these solutions in place, running new cables is going to be messy. You can either try to hide it by running it along the exterior walls or ceiling of your house, or drill holes in the wall every few feet, pass a new cable through, and then patch up the holes.

Side note: cat 5 vs. cat 6

Your internet speed will be bottlenecked by the slowest link in the network. Be careful it isn't your cables!

There are two flavors of Ethernet cable. Category 5 is cheaper, but can only support speeds of 100 Mbps. Category 6 is slightly more expensive but you will need it to get full gigabit speeds.

The Ethernet cables that come with the products you buy may be Cat 5 or 6. Be careful to check which one you are using (it should be written in small print on the outside of the cable).

DHCP

To load google.com, your computer looks up the IP address for Google and sends packets to it. So far so good, but how does Google send packets back? Each client on the network needs a unique local IP address. The router will translate between an open port to Google, say, port 44982, and a local IP address, say, 192.168.0.125, and send packets it receives from the broader Internet on port 44982 to the client with that IP address.

What happens if two clients on your network try to claim the same local IP address? That would be bad. Generally you set up a DHCP server to figure this out. When your phone connects to a wifi network it sends out a packet that says basically "I need an IP address." If a DHCP server is listening anywhere on the network it will find an empty IP address slot and send it back to the phone.2 The phone can then use that IP address.

Generally speaking, a consumer wireless router has three components:

  • wireless radios, that broadcast a network SSID and send packets to and from wireless clients.
  • an Ethernet switch that can split an incoming Internet connection into four or more out ports. Generally this has one WAN port (that connects to your modem/ISP) and four LAN ports (that connect to local devices on your network)
  • a DHCP server.

You can buy products that offer each of these independently - a four way switch without a radio or DHCP server will cost you about $15. But this is a convenient bundle for home networks.

If your network contains multiple switches or multiple routers you need to think about which of these devices will be giving out DHCP.

Two Routers, Too Furious

At this point my network had one router in the bedroom and one router upstairs in the living room, via an ungainly cable up the stairs. So I had good coverage in every room, and the Roku hooked up via Ethernet to the living room router, but this setup still had a few problems. I didn't have the office wired up, and the NAS only worked when you were connected to the living room router.

Furthermore, I kept running into issues where I would walk from the living room to the bedroom or vice versa but my phone/laptop would stay connected to the router in the room I was just in. Because that router was outside its normal "range", I would get more latency and dropped packets than usual, which was frustrating.

How to diagnose and measure this problem

On your laptop, hold down Option when you click the wifi button, and you'll get an extended menu that looks like this.

The key value there is the RSSI parameter, which measures the signal quality from your client to the router. This is currently at -46, a quite good value. Lower than -65 and your connection quality will start to get dicey - you will see lower bandwidth and higher latency and dropped packets.

Apple devices will hang on to the router they are currently connected to until the RSSI gets to -75 or worse, which is a very low value. This is explained in gory detail on this page. Because router coverage areas are supposed to overlap a little bit, this means the connection will have to get very bad before your phone or laptop will start looking for a new radio.

Adjust the power

Generally this means that you don't want the coverage area for the router to reach to the center of the coverage area for the other router, if you can help it. If the coverage areas don't overlap that much, clients will roam to the closest router, which will improve the connection.

You can adjust the coverage area either by physically moving the router or by lowering the power for the radios (which you may be able to do in the admin panel for the router).

If neither of these works, as a last ditch attempt you can give your routers different network names. But this makes it more difficult to keep a connection when you roam from one router to the other.

Ethernet Over... Coax?

I had not managed to get a fixed connection to the office, which would have required snaking a Ethernet cable over at least two doorways and three walls. However, I heard recently about a new technology called MoCA (multimedia over coax), which makes it possible to send an Ethernet signal over the coax line from the garage to the office. I bought a MoCA adapter for each end of the connection - about $160 in total - and wired it up and... it worked like a charm!

Moca ethernet over coax connector in office

The latency is slightly higher than traditional Ethernet, but only by a few milliseconds, and the bandwidth is not as high as a normal wired connection but it's fine - I am still glad to be able to avoid a wireless connection in that room.

This change let me move my NAS into the office as well, which I'm quite happy about.

Letting Everything Talk to Each Other

At this point I had a $15 unmanaged switch in the garage that received a connection from the Sonic Fiber router, and sent it to three places - the bedroom, the living room and my office. However, the fact that it was unmanaged meant that each location requested a public IP address and DHCP from Sonic. Sonic was not happy with this arrangement - there is a limit of 8 devices per account that are stored in a table mapping a MAC address to an IP address, and after this you need to call in to have the table cleared out. This design also meant that the clients on my network couldn't talk to each other - I couldn't access the NAS unless I was connected to the living room router.

The solution was to upgrade to a "managed" switch in the garage that could give out DHCP. You can buy one that is essentially a wifi router without the radio for about $60. This has the same dashboard interface as your router does and can give out DHCP.

Once this switch was in place, I needed to update the routers to stop giving out DHCP (or put them in "pass through mode") so only a single device on the network was assigning IP addresses. I watched the routers and NAS connect, then assigned static IP's on the local network to each one. It's important to do this before you set them in pass-through mode so you can still access them and tweak their settings.

You should be able to find instructions on pass-through mode or "disable DHCP" for your router online. You may need to change the IP address for the router to match the static IP you gave out in the previous paragraph.

That's it

I finally have a network that supported everything I want to do with it! I can never move now.

Garage router setup

I hope this post was helpful. I think the most important thing to realize is that if you haven't done this in a few years, or your only experience is with consumer grade routers, there are other tools/products you can buy to make your network better.

If you are interested in this space, or interested in improving your office network along these lines, I'm working with a company that is making this drop dead easy to accomplish. Get in touch!

1. I posted on the sonic.net forums to get help several times. Dane Jasper, the Sonic CEO who's active on the forums, responded to most of my questions with "you should just use Eero." I love that he is on the forums but Eero is just not great for what I'm trying to do.

2. I'm simplifying - there are two roundtrips, not one - but the details are really not that important.

Liked what you read? I am available for hire.

To Predict If You’ll Like a Beer, Look at the Hops

Posted on

Generally if you name a food or drink, people know whether they like it or not. It is rare for someone to drink a merlot, or try pizza from a new restaurant — toasted bread, melted cheese, tomato sauce and toppings - and be wildly surprised at their reaction to the taste.

I can't quite figure that out for pale ales though. Some pale ales and IPA's had flavors I really liked, and some had flavors I really disliked. I had a tough time predicting which ones I would like and not like.

I had some suspicions - I didn't think I liked beers with much higher ABV than normal or beers that had citrus in them. But I also liked some beers with high ABV and one of my favorite "everyone has it" beers - Sierra Nevada - describes itself as "pine and citrus," so that wasn't quite right.

Anyway, I decided to be somewhat rigorous about this and order a few different types of beers from the bottle shop, and then figure out what I liked or didn't like about them. It turns out the key is the hops - there are some hop varieties (Cascade, Chinook, Noble) that I like a lot, and other hop varieties (Citra, Galaxy, Enigma, others) that I don't at all. If the hop description mentions passion fruit, I probably won't like it. Other than that, I can keep lists.

This is both satisfying - I can predict which beers I will like and not like, now — and frustrating. Why is this so difficult for consumers to figure out? Why does the category definition of "pale ale" include so much stuff? Like imagine if you ordered a "cheese pizza", and sometimes it would come with anchovies and sometimes with pineapple, and sometimes with nothing. People would demand better words to describe the differences between the things.

If you have ideas or answers, I would love to hear from you.

Liked what you read? I am available for hire.

If someone asks if you have any questions, ask a question

Posted on

Let's revisit one of the most humiliating (and expensive) moments of my life. It happened a decade ago and even today I cringe and seethe when I think about it.

I was one of 25 finalists for a $20,000 scholarship in my junior year of college. The last step was an hour long interview with three faculty members. I wrote down a list of every single question I thought they would ask - why do you want this scholarship, why you, etc - and rehearsed answers, recording myself, for a week straight. The interview came and went and I thought I did pretty well!

Fast forward a week and I got an email that I was not going to be offered a scholarship. Only two other students out of 25 were rejected. I was dumbfounded. There was no way I should have failed this test. I started thinking back to the interview. Some answers stand out as opportunities to improve - I could still tell you exactly what they are. There's one answer that I really wish I could get back.

At the end of the interview they asked "do you have any questions for us?" By this point I'd done so much research into the scholarship that I couldn't think of anything, and said "No." The interview ended.

But think about it from their perspective. I'd just spent an hour talking about myself; what does it show when I refuse to ask any questions? Not only am I denying them an opportunity for them to talk, I appear very incurious about the program itself. They didn't know that I had been quizzing myself on every aspect of the program for a week. I should have asked about anything — literally anything — and given them a chance to talk.

The funny thing was I had actually asked some of the people who had gone through the interview what they had been asked about. "Ask a question at the end" was both so obvious to them - pretty much every successful candidate had done finance interviews, where cultural signals are more important; I hadn't - and so oblivious to me that it hadn't even come up to people who wanted to help me succeed.

From that point on I made a point to always ask a question when someone asks if I have any questions. Ask anything. Even asking "what did you have for lunch" is better than asking nothing; the interviewer might start talking about whether the company pays for lunch, whether it's any good. My standby question is "what did you do yesterday" - it has a unique answer for each interviewer and reveals how people spend their time (vs. how they say they spend their time).

Finally, "person fails interview because interviewer expects to see cultural signal and interviewee does not broadcast cultural signal" is a common failure mode. Think about someone who wears a suit to a tech interview. Some organizations only want to hire people who can utter the secret words, and that's their choice.

But if your goal is to cast a wide net - I am looking at you, tech companies that put up billboards championing your commitment to diversity - and you have a candidate without a traditional background, maybe make a list of every reason you've used to reject a nontraditional candidate in the past and then email that to the candidate in advance of the interview - "wear a dress shirt and jeans," for example.1 You won't get everything, but it's a good start. (You can try to get your interviewers to discard the cultural signals but that's difficult and it might show up in their feedback without them realizing it.) Note that career services departments at good schools are already doing this for their students; what you are doing is leveling the playing field.

1 This is not a new critique by any means, people have made it about the SAT for some time - if the tests quiz applicants on vocabulary and grammar that are more commonly used in white households than nonwhite households, than identical students with identical aptitudes will score differently, which seems unfair.

Liked what you read? I am available for hire.

Using a Bernzomatic TS8000 Kitchen Torch to Sear Meat

Posted on

For a company trying to sell and explain a product, a lot of this information was amazingly difficult to find so I wrote this.

For the rest of this we'll assume you're cooking a steak but the same advice applies to most other meats/cuts.

Why Do You Want A Torch to Sear Meat

To get a good texture on a steak, you need to get the outside of the steak really hot. If you keep the steak on the heat source for too long, though, it will start to cook the inside too much, and you'll end up with a well done steak.

A torch helps with this problem by making it possible to sear the outside of the steak very quickly which means the outside can get a nice sear while the inside stays at medium rare.

What to Get

You want a torch that can char the surface of a steak in under a minute. Most of the torches can't actually get that hot!

I used a butane crème brûlée torch for years and would get frustrated my steaks didn't look like the ones on Serious Eats, unless I torched them for 5-6 minutes. The torch wasn't hot enough. Using a cast iron pan by itself, even on the highest setting is also not likely to get hot enough. If your steak does not look as black as the one in Serious Eats the outside is not getting hot enough.

I read the Wirecutter guide and got the Bernzomatic TS8000. This torch is definitely hot enough; with propane it's about 3x as hot as my crème brûlée torch. I used it tonight to sear a tri tip. It is so much better.

Is that it? (No, you need fuel)

No. The Bernzomatic torch is just a torch; you still need to buy fuel. The TS8000 works with either butane or propane. You don't want butane though, you want propane, because it gets 50% hotter than butane does.

Every photo on the Internet of both the TS8000 and propane fuel tanks either shows them connected or with the caps on. You need a propane tank that can connect to this, on the bottom of the TS8000.

The top of your propane canister should look like this, with the cap off:

Bernzomatic sells a compatible 14 oz propane tank. You can also use the green Coleman camping gas tanks. The advantage of the Coleman tank is it is cheap and your grocery/hardware store probably has them. It weighs 16 ounces full, which sounds heavy but, it's fine, it's what I used.

I tried to find a small tank - 5-10 ounces - but didn't find anything promising. You'll probably get something in the 12-16 ounce range. Just put it on the end of the torch and sear with the whole tank on the end of the torch.

Summary

  • Get the Bernzomatic TS8000, it is one of the few torches that gets hot enough

  • Use propane, not butane, with it

  • Coleman camping gas tanks are fine

That's it. Here is the manual for the TS8000. Happy searing!

Liked what you read? I am available for hire.

Phone Number for SFMTA Temporary Sign Office

Posted on

The phone number for the SFMTA Temporary Sign Office is very difficult to find. The SFMTA Temporary Sign web page directs you to 311. 311 does not know the right procedures for the Temporary Sign Office.

The email address on the website is also slow to get back to requests. The Temporary Sign department address listed on the website, at 1508 Bancroft Avenue, is not open to the public — it's just a locked door.

To contact the Temporary Sign Office, call 415-550-2716. This is the direct line to the department. I reached someone in under a minute.

If your event is more than 90 days in the future, don't expect an update. They don't start processing signage applications until 90 days before the event.

Here's a photo of my large son outside of the SFMTA Temporary Sign Office, where I did not find anyone to speak with, but I found the phone number that got me the right phone number to get someone to give me an update on my application.

Liked what you read? I am available for hire.

Using an AWS Aurora Postgres Database as a Source for Database Manager Service

Posted on

Say you have a Aurora RDS PostgreSQL database that you want to use as the source database for Amazon Database Manager Service.

The documentation is unclear on this point so here you go: you can't use an Aurora RDS PostgreSQL database as the source database because Aurora doesn't support the concept of replication slots, which is how Amazon DMS migrates data from one database to another.

Better luck with other migration tools!

Liked what you read? I am available for hire.

Going Solo, Successfully

Posted on

Three years ago I quit my job and started consulting full time. It's worked out really well. I get to spend more time doing things I like to do and I've been able to deliver great products for clients. I wanted to go over some tips for starting a successful consulting business.

  • Charge more - Everyone says it and it's true. I started out charging a monthly rate that was close to my full time salary / 12. This is not a good idea because you have overhead that your employer is no longer covering - health care probably the biggest one, you don't have paid vacations, there may be unpaid downtime between contracts and also companies might not pay you. You need to be charging a lot more to just break even.

    I dread "what's your rate" conversations every time and they haven't gotten easier. Before I quote my rate I reread the details of the High Tech Employee Antitrust case to pump myself up - it reminds me that I'm negotiating with a company that couldn't care less really and I am the only one who's going to stand up for myself. If you think you don't need the extra money - get it anyway, and then donate more to charities at the end of the year/buy CD's/put it in the stock market/give it to the government. Amazon just made $11 billion and paid $0 in taxes; you are going to spend an additional dollar better than Amazon's executives will.

    If you are not sure how much to charge, quote each new client more than the last. Your quote is often a signal of your quality so it's not even really the case that demand slopes downwards as you charge more.

    If you are working with a client and they are very happy with your work and want to extend your contract consider asking for a higher rate. "Now that you know how good I am," etc.

  • Get the money - Signed contracts, work performed, don't matter until the money hits your bank account. I learned this the hard way. If a company is going under your invoices are worthless. You can hold onto the company IP but that's probably also worthless. You can sue but at the end you will win a judgment that you can collect money from a company that doesn't have any to pay you.

    Try to get as much of the contract as you can paid up front - I generally ask for half or more up front. If a company offers Net 30 ask if you can shorten it to Net 5 or 10 or submit your invoices in advance. Submit invoices on time - it's a very costly mistake and you won't learn its importance until it's too late.

    Try as hard as you can to figure out the financial health of the company - if you can do your homework in the press or ask questions to your primary point of contact, like how much cash they are burning, how many months of runway do you have. If a company is not forthcoming with this information it's a red flag that they may not be able to pay you.

    If you see any red flags - the company wants to cut the contract short, people start leaving, company suddenly cuts back on perks - tell your contact that you need to be paid upfront or you are not going to work anymore. If they push back on this they may not have the cash to pay you at all. It's a crappy situation but better to cut your losses than to work for someone that can't actually pay.

  • Don't charge by the hour - I have never actually done this so I can't speak to how bad it is but don't do this. You don't want a client to cut you loose at 3pm and suddenly you lose three hours you were counting on. Charge per week.

  • Get a lawyer - Get a lawyer to review every contract you sign. Read through them, flag concerning things to the lawyer. They will suggest language. Send the language to the company. You are not being difficult when you do this, the company does this all the time. Lawyers are expensive, expect to pay north of $400 per hour and contract review can take 30-60 minutes. This money is worth it.

    A good clause to try to push for is limitation of liability. You don't want to be in a situation where $2 million of damages occurred or a high value client left the company because of an error you pushed and the company is suddenly coming after everything you own. Similarly the company may want to protect against you trying to sue them for a high amount of damages to your reputation, future business etc. Limiting the total liability to the size of the contract, or a multiple of the size of the contract - on both sides - can be a good move.

  • Register as a Company - Consult with the lawyer you hired on what kind of company you want to be. Generally the more "company-like" you are the harder it is for companies to try to take your personal assets. I don't have employees or shareholders so I have a single member LLC that is disregarded for tax purposes — read this description from the IRS. Sometimes businesses are confused what this means when I tell them or try to sign up for things. Still, it is a good fit for me. It may not be for you - I am not a lawyer, you should talk with one, learn the tradeoffs and see what makes sense for your business.

  • Make Sure Contracts Are Signed With the Company - The contracts you sign should be between the client you are working with and your company NOT between the client and you personally. Discuss this with your lawyer.

  • Get an accountant - As a small business a lot of stuff is tax deductible - a home office, client travel, for example, even if it's just across town - and you want to make sure you are getting ~35% off on everything that you can. An accountant will help you with this.

  • Market yourself - Not necessarily ads or sponsorships, but: everyone you've worked with full time should know they can hire you now. If they don't then reach out to people and let them know. Put up a website that engineers can send to their boss. My website isn't fancy but it is effective. Order business cards - VistaPrint is garbage, order from moo.com. If you have a website or open source projects put a note at the bottom advertising that you're available for hire, like the one at the bottom of this post.

  • Set up separate accounts for everything - Open separate accounts for your business. Get a business credit card or just a separate cash back card on your personal account. I don't have a checking account registered for the business but I opened a second checking account that I call my "business account". Clients pay into that account and I pay the business credit card out of that account. I even have a separate Clipper card that I use for business travel.

    There are two reasons for this. It makes tax accounting a lot easier. I know that every transaction on the business Clipper card is work travel and can be expensed; I don't have to try to remember what I was doing when I paid $2 to SamTrans at 5:34pm on July 27.

    Second, if you don't keep good records for the business - if you "commingle" funds between your personal life and the business - it makes it much easier for clients to go after your personal assets, what's called "piercing the veil." Separate accounts (and discipline about transfers!) make it much easier to argue that your business income and spending and personal income and spending are separate even if you don't necessarily have the legal structures to back them up.

    I also set up a new Github account for every company I work with. This avoids any issues with emails going to the wrong place, or the need to grant/revoke permissions to any 3rd party tools a company uses. I use github.com/kevinburke/swish to swap SSH settings between my Github accounts:

    $ cat $(which notion-github)
    #!/usr/bin/env bash
    ${GOPATH}/bin/swish --identity-file ${HOME}/.ssh/github_notion_ed25519 --user kevinburkenotion

  • Balancing multiple clients: If you can do this or do things like charge retainers, great. I find it really hard to switch contexts so I work with one client at a time and treat it as a full time job. Do what works for you.

  • Give back to the tools that make you successful - I give a percentage of my earnings every year to support software tools that help me do my job - iTerm2, Vim, Go, Postgres, Node.js, Python, nginx, various other open source projects. You should consider doing this too. (If you are an open source maintainer reading this - tell me how I can pay you!!)

Liked what you read? I am available for hire.

Improving Bike+Bus Routes through Belmont Village

Posted on

Belmont is evaluating pedestrian and bike improvements to Ralston Avenue. These improvements build on the Pedestrian and Bicycle Master Plan.

There is pretty good cycle access all the way from the eastern tip of Foster City to the Belmont Caltrain station.

However, the cycling access from the Caltrain station to Safeway, NDNU or Carlmont Village falls down a little bit. There is a huge intersection at Ralston and ECR with no bike paths. The bike plan says cyclists should ride on Emmett past City Hall and rejoin Ralston further west, but there are no bike lanes in this area.

Emmett has no bike lanes and parking on both sides.

If you are riding east it is difficult to get from Emmett to the Caltrain station - cars can make a left turn across two lanes of ECR into fast moving traffic and then merge right, but this is dicey on a bike. I usually ride on the sidewalk past the Panda Express to the intersection. The sidewalk is narrow and passes the ECR Rapid bus stop.

The bike plan (page 72) calls for a HAWK crossing at ECR and Emmett to allow bikes to get across, then a bike lane on the east side of ECR. HAWK intersections are confusing for drivers, and no guarantee for ped/bike safety - your author recently spotted an instruction sign for drivers on the side of the road near a HAWK beacon on El Camino Real in Redwood City.

It would also require cyclists to make a dangerous crossing across ECR that no one currently attempts, cross the driveway for the parking lot, and add an additional stop that reduces vehicle/SamTrans ECR bus levels of service along El Camino Real.

How else could we get bikes from Caltrain to Emmett? One idea would be to have a two-way cycle track on the south side of ECR where there is currently a bus stop and a lane of parked cars. The bus would stop in the lane on ECR, like in this photo - we wouldn't have enough room for a bench to the left of the cycle track, but this should give the basic idea.

This would do a few different things, all of them good.

First, it does not introduce an additional stopping point for cars and SamTrans buses along El Camino Real, which means those roads will have a higher level of service.

Second, it would remove the need for a dangerous HAWK crossing and add a protected path on a dangerous section of road for cyclists. Particularly if you are heading east toward Caltrain, there are currently no good options for getting to the station.

Third, it would make the bus faster. Bus speeds are a problem for SamTrans that led to the introduction of the "ECR Rapid" bus route. Part of the problem are stops that require the bus to pull over, load passengers, then wait for a gap in traffic before merging back in. In-lane boarding would allow the bus to continue its journey without needing to merge back into traffic, saving up to 40 seconds per trip.

Fourth, it would likely be faster, by allowing bikes to piggy back on the existing pedestrian crossing at Ralston and ECR. HAWK beacons require all traffic to wait for the beacon to stop traffic, which can take thirty seconds or more.

Fifth, it could permit installation of an ADA compliant bus stop. The current bus stop offloads onto a brick sidewalk on a slope, intermixed with trees. A treatment in the cycle track segment could include a level boarding space for wheelchairs.

There are some problems, notably that there is not much room for this. Curb-to-median, this stretch of El Camino Real is 32 feet. The preferable alternative would be a bike lane protected by a bus island, like this one on Church St. in San Francisco. We can't do this because the recommended minimum width for bus islands is 8 feet, space that we don't have.

There is a one foot gutter next to the median and the car lanes are 10 feet each, giving us 11 feet for a 4 foot cycle track in each direction, plus a small buffer zone between the track and the car lane, which could include a railing for additional safety. The entire cycle track/buffer would be raised to the level of the current curb.

If you moved the limit line on ECR on the south side of Ralston back ten feet, you could add a treatment like this for crossing ECR on your bike:

Which would result in a bike route through town like this, in both directions.

What do you think? I am not super happy with this design, but I'm not sure how else cyclists should be expected to get from Emmett St. to the Caltrain station. Moving the bus boarding in-lane would be a big win.

There is a meeting October 10, 6pm at City Hall where Belmont Public Works will be revealing plans for the HAWK beacon and other bike improvements. Please show up and provide your feedback.

Liked what you read? I am available for hire.

Why is Caltrain spending $2 billion on electrification & closing SF on the weekends?

Posted on

Caltrain is spending $2 billion to electrify the entire railway from SF to Tamien, and buy new train cars. This is expensive and painful - weekend headways are 90 minutes, and the 4th and King/22nd St. stations won't get trains on Saturdays and Sundays for about six months. Why is Caltrain doing this? There are a few reasons.

  • Faster trains. With the diesel trains the only car that can accelerate is the locomotive. With the new cars, every car will be capable of pulling electricity from the overhead line - they are called "electric multiple units" because you have multiple cars that can each accelerate independently. This means the trains can accelerate and decelerate much more quickly a diesel engine can — the riding experience will be similar to riding BART.

  • Faster boarding. The old Caltrain "Gallery" cars, representing most of the fleet, have a floor height of 50 inches, which means passengers and bikes need to ascend/descend multiple stairs to embark/disembark. The new train cars (and the Bombardier cars) have a floor height of 25 inches, which means it's a lot faster for passengers to get on and off the trains. These combine to mean...

  • Shorter trip times. If you can accelerate and decelerate more quickly, and you can get passengers on and off more quickly, you can get from SF to Diridon faster, or you can make more stops in the same amount of time. Further, if you have shorter trip times, you can have...

  • More frequent service. If the trains can get from end to end faster, you can run more trains. Further, since every train is capable of being the "engine", it's less painful to run shorter trains, more frequently during off-peak times. Think "not having to worry about catching the Giants Special, because the trains run every 15 minutes no matter what."

  • Less respiratory disease. Diesel trains burn, well, diesel, in order to operate, which means the air gets pretty dirty. Pollution is a leading cause of respiratory diseases like asthma and bronchitis, not to mention permanently accelerated aging of the lungs, and killing plants along the railway. Burning diesel also contributes to the slow warming of the planet, which causes more fires and flooding.

    Electrification means the trains need less energy to run and also that energy burns cleaner, which should lead to cleaner air and healthier communities.

Those are the main reasons to spend $2 billion on electrifying Caltrain. You might be saying, the US is not great at delivering infrastructure projects for a reasonable budget, is $2 billion a reasonable amount to spend? The answer is largely, Yes. The San Mateo County grand jury just investigated the contract award and decided that Caltrain was implementing best practices in both the award of the contract and the oversight.

We are not out of the woods yet. Last year the California Legislature passed a twelve-cent-per-gallon tax on gasoline, which is being used to fund road repairs and transit improvements up and down the state. Caltrain is depending on this revenue to help fund the electrification project. This fall, voters will vote on whether that tax should be repealed, which would hurt Caltrain's ability to fund electrification on time and on budget. Please vote no on Prop 6 this fall.

Liked what you read? I am available for hire.