Max Schireson's blog

I love free software, and Richard Stallman has done much to advance it, but this time he’s really gone too far.

From his blog:

“Steve Jobs, the pioneer of the computer as a jail made cool, designed to sever fools from their freedom, has died.

As Chicago Mayor Harold Washington said of the corrupt former Mayor Daley, “I’m not glad he’s dead, but I’m glad he’s gone.” Nobody deserves to have to die – not Jobs, not Mr. Bill, not even people guilty of bigger evils than theirs. But we all deserve the end of Jobs’ malign influence on people’s computing.

Unfortunately, that influence continues despite his absence. We can only hope his successors, as they attempt to carry on his legacy, will be less effective.”

I can’t speak on behalf of the entire open source community, but I can say that Richard Stallman certainly does not on this point.

My quick take:

Apple, under Jobs leadership, built some great products. They weren’t free (like beer or speech), but they moved the industry forward. Would it have been even better if they were free? Sure. But that’s a different world. I’m glad that folks have the freedom to make great software which is free and great software which is not. “Freedom” can be its own tyranny.

Love to hear your thoughts…

– Max

A room is 30 feet long, 12 feet wide, and 12 feet high.

An ant is one foot from the ceiling in the middle of the 12 foot long wall on the north end of the room.

A drop of honey is in the middle of the 12 foot long wall on the south end of the room, one foot from the floor.

How far does the ant have to walk to get the honey.

Hint: if the obvious path were the shortest, I wouldn’t have bothered to post the problem.

Have fun!

– Max

I’ve gotten a lot of questions from press, analysts, and customers about Oracle’s NoSQL offering. Our booth at Oracle Open World has been mobbed (next year we need a bigger booth – this year we had to make do with spilling over about 8 feet into the walkway on either side of our booth, sorry for the blockage for anyone trying to get past us!). I thought I’d share some further thoughts on Oracle’s now-announced offering here.

Disclaimer: I work for 10gen, the company that sponsors MongoDB, so I have some obvious bias. That said, I strongly believe that solid offerings from big players will help the market overall. With that bias disclosed, I have done my best to be fair and intellectually honest in my analysis of their offering.

About the offering

Personally, I think the Oracle offering is reasonable. Their approach to consistency looks sensible, as do their replication and sharding designs. It looks like it will be available open source, so I expect it will get a reasonable amount of adoption. I don’t think it breaks new ground, but if you want a key-value store from Oracle, it looks like it will do the job. My prediction is that it won’t crack the top 3 in NoSQL, but it looks solid and with Oracle behind it I’d expect it to wind up somewhere around # 5 or so in a very competitive market. This is just based on looking at their white papers; final judgement will have to wait until the product is released.

One plus for their offering is that it is based very heavily on BerkeleyDB. This technology has been around for at least 15 years, so it should be pretty stable. The main addition appears to be hash-based sharing; this is the part that customers need to ensure is stable before adopting it, but I’d expect the rest of it to be quite solid.

On criticism for flip-flopping

There have been a lot of harsh words about Oracle’s switch from writing a white paper on “Debunking the NoSQL Hype” in May to announcing a NoSQL offering in October. I wish it were otherwise, but welcome to the world of high-tech marketing. When a big company feels pain in the market and doesn’t have a product, they have their marketing guys write something that says the market isn’t important. You can be grumpy that the industry works that way and think whoever decided to do that is a hypocrite, but it really has no bearing on Oracle’s commitment to the product now that they have built it. Time to move on and evaluate the product on its merits.

So, congrats to Oracle on entering an exciting and growing market segment with what looks to be a solid offering. I look forward to seeing how it evolves over its next few releases and to competing with them in marketplace!

– Max

Recently there have been rumors of Oracle introducing a NoSQL database at Open World next week. As a 9-year Oracle veteran as well as an 8-year veteran of the alternative database world (most recently as President of 10gen, the sponsor of MongoDB), this is an exciting development for me.

I don’t think NoSQL means the end of SQL or relational – its one of the reasons I don’t love the name “NoSQL”. I do, however, think that relational is not the answer to everything. When I left Oracle eight years ago to work on alternatives to relational database, I felt that there was space for an alternative that was more agile for developers and more scalable for deployment on a large number of commodity servers. The last eight years with the emergence of web 2.0 and cloud deployment have only strengthened my belief that relational is not the answer to all questions.

How do I feel about Oracle introducing a “NoSQL” database”? I think it makes all the sense in the world. Oracle already has acquired a variety of database products ranging from the acquisition of RDB from DEC to Sleepycat and TimesTen to MySQL (acquired with SUN). Much of the innovation in the database world right now is focused on scalable non-relational databases, most of which are open source and  which tend to be lumped under the “NoSQL” banner. The innovation is focused there for three reasons:
- Customers want horizontal scalability to support deployment on clusters of commodity servers, often in the cloud
- Customers want greater agility in their database to accomodate data whose structure is changing or heterogeneous
- Customers want easier development

In my opinion this is a good thing for alternative database vendors. Competition is already thriving in the sector and I don’t think one more competitor, even one as large as Oracle, will alter the  dynamics dramatically. But many customers will take Oracle’s arrival in the space as a sign that this trend is significant and it is a space they should look at. If Oracle’s offering is strong, we may lose some market share to them, but their presence will make it a bigger market.

In my time at Oracle, I found Larry Ellison to have a great sense of what markets were important and to be a fierce competitor. In some markets Oracle has crushed competitors (eg, RDBMS vs Sybase, Informix, and Ingres); in others it has struggled for a time and eventually succeeded through acquisition (eg, enterprise apps vs Peoplesoft, Siebel, and SAP); in some its presence has barely been felt (eg, SaaS vs Salesforce and NetSuite). How will Oracle’s entry into the NoSQL market fare? I don’t know, but I’d love to hear your thoughts.

Oracle, welcome to what in my opinion is the most exciting part of the database market!

– Max

It is not often that service is bad enough to be worth writing about. Tonight’s bartender in London was something special.

After dinner a few of us ordered drinks in the hotel bar. The menu was somewhat limited (after hours room service menu), but one of the cocktails they offered was a Martini. The only Gin they had was Tanqueray.

I ordered a Tanqueray Martini, straight up, with olives. Not particularly complex. They also had only one vodka; my colleague ordered an Absolut Martini with blue cheese olives. Surprised when he took the order I asked if they had blue cheese olives. He confirmed, and I asked to change my order to a Tanqueray Martini with blue cheese olives.

You are probably thinking that I am grumpy that I didn’t get my blue cheese olives. If only that was the biggest problem with my order. Even if I’d gotten a Martini on the rocks with regular olives, it wouldn’t have been that noteworthy.

20 minutes later, however, what arrived was lukewarm, had no olives, and didn’t smell like gin. It also wasn’t quite clear, slightly yellowish. After a bit of conversation we determined that what we’d gotten were two glass of warm vermouth. When we said we didn’t want them, he said ok, but that we should pay anyway. He even brought out the bottle of Martini brand vermouth to support his case that he’d made us exactly what we asked for. Incredibly, all the discussion of olives, Tanqueray, and Absolut had been completely ignored, as had the traditional definition of a Martini.

Even if you have no idea how to make a Martini (in which case I you need to find a different line of work), what on earth is going through your head when there is all of this talk of Tanqueray and Absolute and blue cheese vs regular olives?

Giving up on getting a Martini, I ordered a Stella (primarily because I’d seen someone else get one correctly). My colleague ordered a Jameson’s and got a Jack Daniels.

Amazing.

– Max

After being ousted as CEO of Yahoo, Carol Bartz said she would remain on the board. Under some circumstances, that might have been a good way to ensure a smooth transition.

However, calling your fellow board members “doofuses”  and claiming they “f***ed me over” in an interview with Fortune probably doesn’t help the interpersonal dynamics on the board. Unsurprisingly, she is leaving the board.

Note to execs: I’m pretty sure boards don’t like being called “doofuses”, nor their behaviors being described with the f-word. Not sure saying you were “screwed over” by “buffoons” would work much better.

I am not an expert on tact or euphemism, but here are some alternatives to consider if you must describe your firing and the individuals who did it and want to remain on the board. ”Held accountable for a stagnant stock price” by “hard-nosed shareholder representatives” might even be a little tough. How about:

• “Encouraged to step back from day to day operations” by “people I trusted to have the company’s best interests at heart”
• “Supported in a decision to move on” by “a strong team I have very much enjoyed working with”
• “Helped to see that new leadership was necessary” by “a number of my key advisors”

– Max

Computer hardware architectures and the performance of various components are co-evolving in a complex way, but one of the results is that “RAM” is becoming less well suited to random access – or at least relatively better optimized for sequential access. In the last decade or so from SDRAM to DDR3, memory transfer rates have improved from 800MB/second for PC100 memory to 17066MB/second for DDR3-2166, or a little better than 21x improvement. Access time, however, has barely improved 2x, from around 60ns to around 30ns. So the relative difference between sequential and random memory access performance has increased more than 10x. Its not quite the random/sequential speed difference of tape, but that’s a very significant change.

What’s going on? I’ll go through memory speeds in the “old days” (I think for everyone based on when they learned to program; for me that’s the early 80s) and how things have evolved more recently. Warning: unless you are really a serious geek, having heard the main point you should probably stop reading here.

I started programming on an Apple II+, first in BASIC then in assembly language (after that I moved to a VAX, programming mostly in C). I still remember on a 6502 A9 would load the accumulator with an immediate value (one byte). Its all a little hazy now, but the length of time to perform an operation had to do in large part with memory accesses. Most immediate operations (operations where the operand was specified as a constant, for example LDA #$AA, which would load the accumulator with the value $AA) ran in two clock cycles. Zero-page addressing (the first 256 bytes of memory) would add an extra cycle, so LDA $AA, which loads the accumulator with whatever value was in memory location $00AA, would run in 3 clock cycles. Absolute addressing, for example LDA $AAAA which loads the accumulator with whatever value is in memory location $AAAA, would run in 4 clock cycles. It didn’t actually cost anything extra to offset those absolute pointers by the X or Y register as long as you didn’t cross a page boundary when you added the offset.

Indirect zero page addressing would take 5 or 6 cycles, and is worth explaining because it was the sensible way to do arrays and pointers: LDA ($AA), Y would find the two byte address stored starting at $00AA, add Y to that, and load the value at that address. In an asymmetry which I transposed far too often (and created many bugs in the process) but makes sense when you only have 256 instructions to play with, indirect addressing with the X register was completely different: the X register was added to the immediate value used for the zero page address before dereferencing. This was useful for tables of pointers but I found the Y register style of indirection useful much more often in my programming.

Anyway, the overhead of accessing memory was anywhere from 1 to 4 CPU clock cycles, which was the equivalent of .5-2 immediate (no memory access) instructions. Memory access times were significant but didn’t absolutely dwarf processing time.

Fast forward to today’s hardware. Much publicized “numbers that every programmer should know:” L1 cache access is .5 ns, L2 cache access is 7 ns, and main memory access is 100 ns. Rather than a 4:1 range in the old days, we now have a 200:1 range. And even worse, in the old days you knew statically (within 1 cycle depending on page boundaries anyway) how much time you would take for memory access; now you don’t know, it depends on whether you get a cache hit.

100 ns is a long time; lets dig in and see what’s behind it. Accessing memory nowadays is a complicated affair, and honestly until I sat down to write this I didn’t have a clear understanding of how that time was spent or how long it would take with different types of memory. Here’s what I found.

Using today’s memory, you don’t just ask for an address. Memory is organized in rows and columns, and before you can access an new row, you first have to precharge, then you strobe the new row address, then you strobe the column address that you want. If you’re looking at memory performance specs, the times for each of these operations are listed as tRP, tRCD, and CL. You have to add these together to determine how many memory I/O bus cycles it takes to read a new address not on the same row as the last read. Careful though, because a memory I/O bus cycle nowadays is twice as long as you think: DDR3-1600 memory actually has an I/O bus speed of 800 MHz, so each cycle is 1.25 ns. A DDR3-1600G rated memory takes 8 cycles each for tRP, tRCD, and CL, so altogether a memory access can happen in 30ns – not as bad as I feared. Well, 100 ns was an order of magnitude figure; its still quite a bit longer than .5 ns for an L1 cache hit and I’m not quite including everything that has to occur in the whole cycle (eg, you need to know you have a cache miss before you even start the process, and there are multiple layers of cache).

If DDR3-1600 memory can return a result in 30ns, how much can we shave the time with DDR3-2133 memory? The answer will surprise you – it might not be faster at all for random access. There is no JEDEC standard for DDR3-2133G; instead it starts at DDR3-2133K as the fastest DDR3-2133 standard. Rather than 8 cycles each for tRP, tRCD, and CL, it costs 11 cycles. Overall a random memory access is 33 cycles. 33 cycles at 1066MHz is about 31ns – no faster at all, and in fact slightly slower. Of course the absence of a JEDEC standard doesn’t mean it isn’t possible – a true weenie will buy his memory and test it with various BIOS settings. I’ve seen tests which show stable operation in some cases with 9/11/9 cycle timings at DDR3-2133, for a total of 29 cycles or just over 27 ns.

That said, DDR3-2133 memory has a significantly higher transfer rate than DDR3-1600: 17066MB/second vs 12800MB/second. Again, you see throughput increasing but not latency. Going back to the PC100 memory, that was common in PCs a decade ago, you see the same patter: an 800MB/second transfer rate based on a 100MHz memory I/O bus and one transfer per cycle (rather than the current two), but tRP, tRCD, and CL were all 2 cycles. A random memory access was 6 cycles of 100MHz, or 60 ns, only twice the access time of today’s high end memory as compared to transfer rates 21 times slower than today.

Where this may be headed in the future will be the subject of another post.

Thanks,

– Max

Stock options are a big part of the startup dream but they are often not well understood, even by senior execs who derive much of their income from stock options. Here’s my attempt to explain the main issues employees should be aware of.

What they are

“Stock options” as typically granted give you the right to buy shares of stock in the future for a price which is determined today. The “strike price” is the price at which you can buy the shares in the future. If in the future the stock is worth more than the strike price, you can make money by “exercising” the options and buying a share of stock for the strike price. For example, your are granted 5,000 shares of stock at $4 per share in a startup. 5 years later, the stock goes public and three years after that it’s run up to $200 per share. You can exercise the option, paying $20,000 to buy 5,000 shares of stock which are worth $1,000,000. Congrats, you’ve made a $980,000 pretax profit, assuming you sell the shares immediately.

Vesting

There is a small but necessary catch: when you are granted your options, they are not “vested”. This means that if you leave the company the week after you join, you lose your stock options. This makes sense; otherwise rather than being an incentive to stay, they’d be an incentive to job-hop as much as possible, collecting options from as many employers as you can. So, how long do you have to stay to keep your options? In most companies, they vest over four years. The most common structure is a “cliff” after one year when 25% of your shares vest, with the remaining shares vesting pro-rata on a monthly basis until you reach four years. Details vary from company to company; some companies vest options over 5 years and some over other periods of time, and not all employers have the cliff.

The cliff is there to protect the company – and all the shareholders, including other employees – from having to give shares to individuals who haven’t made meaningful contributions to the company

Dilution

Why should you care about whether that guy who got fired after six months walked away with any options or not?  Because those options “dilute” your ownership of the company. Remember each share represents a piece of ownership of the company. The more shares there are, the less value each one represents. Lets say when you join the startup and get 5,000 shares, there are 25,000,000 total shares outstanding. You own .02% – two basis points – of the company. If the company issues another 25,000,000 options or shares over the intervening five years so there are  50,000,000 shares at the IPO (typically either as part of fundraising including an IPO or to hire employees), you’re left with .01% – one basis point or half of your original percentage. You have had 50% dilution. You now make half as much for the same company value.

That said, dilution is not necessarily bad. The reason the board approves any dilutive transaction (raising money, buying a company, giving out stock options) is that they believe it will make the shares worth more. If your company raises a lot of money, you may own a smaller percentage, but the hope is that the presence of that cash allows the company to execute a strategy which enhances the value of the enterprise enough to more than compensate for the dilution and the price per share goes up. For a given transaction (raising $10 million) the less dilutive it is the better, but raising $15 million may be more dilutive than raising $10 million while increasing the value of each existing share.

Ownership percentage

This brings us to the number which is much more important (though it is less impressive sounding) than the number of shares – what portion of the company do you own. This is often measured in percentage terms, which I think is unfortunate because very few employees other than founders wind up with one percent or even half a percent, so you’re often talking about tiny fractions, which is irritating. I think it is more useful to measure it in “basis points” – hundredths of a percent. Regardless of units, this is the number that matters. Why?

Lets say company A and company B are both, after lots of hard work, worth $10 billion (similar to Red Hat, for example). Long ago Albert went to work at company A and Bob went to work at company B. Albert was disappointed that he only got 5,000 options, and they were granted at a price of $4 each. Bob was very happy – he was granted 50,000 options at only 20 cents each. Who got the better deal? It depends. Lets say company A had 25,000,000 shares outstanding, and company B had 500,000,000 shares outstanding. After many years and 50% dilution in each case, company A has 50,000,000 shares outstanding so they are worth $200 each and Albert has made a profit of $980,000 on his options ($1 million value minus $20,000 exercise cost). Company B has 1 billion shares outstanding, so they are worth $10 each. Bob’s options net him a profit of $9.80 each, for a total profit of $490,000. So while Bob had more options at a lower strike price, he made less money when his company achieved the same outcome.

This becomes clear when you look at ownership percentage. Albert had 2 basis points, Bob had one. Even though it was less shares, Albert had more stock in the only way that matters.

How many shares outstanding is “normal”? At some level the number is totally arbitrary, but many VC funded companies tend to stay in a similar range which varies based on stage. As a company goes through more rounds of funding and hires more employees, it will tend to issue more shares. A “normal” early stage startup might have 25-50 million shares outstanding. A normal mid-stage (significant revenue and multiple funding rounds, lots of employees with a full exec team in place) might have 50-100 million shares outstanding. Late stage companies that are ready to IPO often have over 100 million shares outstanding. In the end the actual number doesn’t matter, what matters is the total number relative to your grant size.

Exercise

I talked briefly about exercising options above. One important thing to keep in mind is that exercising your options costs money. Depending on the strike price and the number of options you have, it might cost quite a bit of money. In many public companies, you can do a “cashless exercise” or “same-day-sale” where you exercise and sell in one transaction and they send you the difference. In most private companies, there is no simple way to do the equivalent. Some private companies allow you to surrender some of the shares you’ve just exercised back to the company at their “fair market value”; read your options agreement to see if this is offered. I’ll talk more about “fair market value” below, but for now I’ll just say that while its great to have this option, it isn’t always the best deal if you have any alternative.

The other really important thing to consider in exercising stock options are taxes, which I will discuss later.

Fair market value

In my opinion, the process by which the “fair market value” of startup stock is determined often produces valuations at which it would be very difficult to find a seller and very easy to find buyers – in other words a value which is often quite a bit lower than most people’s intuitive definition of  market value. The term “fair market value” in this context has a very specific meaning to the IRS, and you should recognize that this technical meaning might not correspond to a price at which it would be a good idea to sell your shares.

Why is the IRS involved and what is going on? Stock option issuance is governed in part by section 409a of the internal revenue code which covers “non-qualified deferred compensation” – compensation workers earn in one year that is paid in a future year, other than contributions to “qualified plans” like 401(k) plans. Stock options present a challenge in determining when the “compensation” is “paid”. Is it “paid” when the option is granted, when it vests, when you exercise the option, or when you sell the shares? One of the factors that the IRS uses to determine this is how the strike price compares to the fair market value. Options granted at below the fair market value cause taxable income, with a penalty, on vesting. This is very bad; you don’t want a tax bill due when your options vest even if you haven’t yet exercised them.

Companies often prefer lower strike prices for the options – this makes the options more attractive to potential employees. The result of this was a de-facto standard to set the “fair market value” for early stage startup options issuance purposes to be equal to 10% of the price investors actually paid for shares (see discussion on classes of stock below).

In the case of startup stock options, they specify that a reasonable valuation method must be used which takes into account all available material information. The types of information they look at are asset values, cash flows, the readily determinable value of comparable entities, and discounts for lack of marketability of the shares. Getting the valuation wrong carries a stiff tax penalty, but if the valuation is done by an independent appraisal, there is a presumption of reasonableness which is rebuttable only upon the IRS showing that the method or its application was “grossly unreasonable”.

Classes of stock

Most startups have both common and preferred shares. The common shares are generally the shares that are owned by the founders and employees and the preferred shares are the shares that are owned by the investors. So what’s the difference? There are often three major differences: liquidation preferences, dividends, and minority shareholder rights plus a variety of other smaller differences. What do these mean and why are they commonly included?

The biggest difference in practice is the liquidation preference, which usually means that the first thing that happens with any proceeds from a sale of the company is that the investors get their money back. The founders/employees only make money when the investors make money. In some financing deals the investors get a 2x or 3x return before anyone else gets paid. Personally I try to avoid those, but they can make the investors willing to do the deal for less shares, so in some situations they can make sense. Investors often ask for a dividend (similar to interest) on their investment, and there are usually some provisions requiring investor consent to sell the company in certain situations.

Employees typically get options on common stock without the dividends or liquidation preference. The shares are therefore not worth quite as much as the preferred shares the investors are buying.

How much are they worth

That is, of course, the big question. If the “fair market value” doesn’t match the price at which you reasonably believe you could find a buyer, how do you about estimating the real world value of your options?

If your company has raised money recently, the price that the investors paid for the preferred shares can be an interesting reference point. My experience has been that a market price (not the official “fair market value”, but what VCs will pay) for common shares is often between 50% and 80% of the price the investors pay for preferred shares. The more likely that the company will be sold at a price low enough that the investors benefit from their preference the greater the difference between the value of the preferred shares and the common shares.

The other thing to keep in mind is that most people don’t have the opportunity to buy preferred shares for the price the VCs are paying. Lots of very sophisticated investors are happy to have the opportunity to invest in top-tier VC funds where the VC’s take 1-2% per year in management fees and 25-30% of the profits. All told, they’re netting around 60% of what they’d net buying the shares directly. So when a VC buys common shares at say 70% of the price of preferred shares, that money is coming from a pension fund or university endowment who is getting 60% or so of the value of that common share. So in effect, a smart investor is indirectly buying your common shares for around the price the VCs pay for preferred.

If there hasn’t been a round recently, valuing your shares is harder. The fair market value might be the closest reference point available, but I have seen cases where it is 30-60% (and occasionally further) below what a rational investor might pay for your shares. If its the only thing you have, you might guess that a market value would be closer to 2x the “fair market value”, though this gap tends to shrink as you get close to an IPO.

Expiration and termination

Options typically expire after 10 years, which means that at that time they need to be exercised or they become worthless. Options also typically terminate 90 days after you leave your job. Even if they are vested, you need to exercise them or lose them at that point. Occasionally this is negotiable, but that is very rare – don’t count on being able to negotiate this, especially after the fact.

The requirement to exercise within 90 days of termination is a very important point to consider in making financial and career plans. If you’re not careful, you can wind up trapped by your stock options; I’ll discuss this below.

Acceleration

Occasionally stock options will have “acceleration” language where they vest early upon certain events, most frequently a change of control. This is an area of asymmetry where senior executives have these provisions much more frequently than rank-and-file employees. There are three main types of acceleration: acceleration on change of control, acceleration on termination, and “double trigger” acceleration which requires both a change of control and your termination to accelerate your vesting. Acceleration can be full (all unvested options) or partial (say, 1 additional year’s vesting or 50% of unvested shares).

In general, I think acceleration language makes sense in two specific cases but doesn’t make sense in most other cases: first, when an executive is hired in large part to sell a company, it provides an appropriate incentive to do so; second when an executive is in a role which is a) likely to be made redundant when the company is sold and b) would be very involved in the sale should it occur it can eliminate some of the personal financial penalty that executive will pay and make it easier for them to focus on doing their job. In this second case, I think a partial acceleration, double trigger is fair. In the first case, full acceleration may be called for, single trigger.

In most other cases, I think executives should get paid when and how everyone else gets paid. Some executives think it is important to get some acceleration on termination. Personally I don’t – I’d rather focus my negotiation on obtaining a favorable deal in the case where I’m successful and stick around for a while.

How many should you get

How many stock options you should get is largely determined by the market and varies quite a bit from position to position. This is a difficult area about which to get information and I’m sure that whatever I say will be controversial, but I’ll do my best to describe the market as I believe it exists today. This is based on my experience at two startups and one large company reviewing around a thousand options grants total, as well as talking to VCs and other executives and reviewing compensation surveys.

First, I’ll talk about how I think about grant sizes, then give some specific guidelines for different positions.

I strongly believe that the most sensible way to think about grant sizes is by dollar value. As discussed above, number of shares doesn’t make sense. While percent of company is better it varies enormously based on stage so it is hard to give broadly applicable advice: 1 basis point (.01 percent) of Google or Oracle is a huge grant for a senior exec but at the same time 1 basis point is a tiny grant for an entry level employee at a raw series-A startup; it might be a fair grant for a mid-level employee at a pre-IPO startup. Dollar value helps account for all of this.

In general for these purposes I would not use the 409a “fair market value”. I would use either a) the value at the most recent round if there was one or b) the price at which you think the company could raise money today if there hasn’t been a round recently.

What I would then look at is the value of the shares you are vesting each year, and how much they are worth if the stock does what the investors would like it to do – increases in value 5-10 times. This is not a guaranteed outcome, nor is it a wild fantasy. What should these amounts be? This varies by job level:

Entry level: expect the annual vesting amount to be comparable to a small annual bonus, likely $500-$2500. Expect the total value if the company does well to be be enough to buy a car, likely $25-50k.

Experienced: most experienced employees will fall in to this range. Expect the annual vesting amount to be comparable to a moderate annual bonus, likely $2500-$10k, and the total value if the company does well to be enough for a down-payment on a silicon valley house or to put a kid through college, likely around $100-200k.

Key management: director-level hires and a handful of very senior individual contributors typically fall into this range. Key early employees often wind up in this range as the company grows. Expect the annual vesting amount to be like a large bonus, likely $10k-40k and the total value if the company does well to be enough to pay off your silicon valley mortgage, likely $500k-$1 million.

Executive: VP, SVP, and CxO (excluding CEO). Expect the annual vesting amount to be a significant fraction of your pay, likely $40-100k+, and the value if the company does well to be $1 million or more.

For those reading this from afar and dreaming of silicon valley riches, this may sound disappointing. Remember, however, that most people will have roughly 10 jobs in a 40 year career in technology. Over the course of that career, 4 successes (less than half) at increasing levels of seniority will pay off your student loans, provide your downpayment, put a kid through college, and eventually pay off your mortgage. Not bad when you consider that you’ll make a salary as well.

What should I ask

You should absolutely ask how many shares are outstanding “fully diluted”. Your employer should be willing to answer this question. I would place no value on the stock options of an employer who would not answer this clearly and unambiguously. “Fully diluted” means not just how many shares are issued today, but how many shares would be outstanding if all shares that have been authorized are issued. This includes employee stock options that have been granted as well shares that have been reserved for issuance to new employees (a stock “pool”; it is normal to set aside a pool with fundraising so that investors can know how many additional shares they should expect to have issued), and other things like warrants that might have been issued in connection with loans.

You should ask how much money the company has in the bank, how fast it is burning cash, and the next time they expect to fundraise. This will influence both how much dilution you should expect and your assessment of the risk of joining the company. Don’t expect to get as precise an answer to this question as the previous one, but in most cases it is reasonable for employees to have a general indication of the company’s cash situation.

You should ask what the strike price has been for recent grants. Nobody will be able to tell you the strike price for a future grant because that is based on the fair market value at the time of the grant (after you start and when the board approves it); I had a friend join a hot gaming company and the strike price increased 3x from the time he accepted the offer to the time he started. Changes are common, though 3x is somewhat unusual.

You should ask if they have a notion of how the company would be valued today, but you might not get an answer. There are three reasons you might not get an answer: one, the company may know a valuation from a very recent round but not be willing to disclose it; two the company may honestly not know what a fair valuation would be; three, they may have some idea but be uncomfortable sharing it for a variety of legitimate reasons. Unless you are joining in a senior executive role where you’ll be involved in fundraising discussions, there’s a good chance you won’t get this question answered, but it can’t hurt to ask.

If you can get a sense of valuation for the company, you can use that to assess the value of your stock options as I described above. If you can’t, I’d use twice the most recent “fair market value” as a reasonable estimate of a current market price when applying my metrics above.

Early exercise 

One feature some stock plans offer is early exercise. With early exercise, you can exercise options before they are vested. The downside of this is that it costs money to exercise them, and there may be tax due upon exercise. The upside is that if the company does well, you may pay far less taxes. Further, you can avoid a situation where you can’t leave your job because you can’t afford the tax bill associated with exercising your stock options (see below where I talk about being trapped by your stock options).

If you do early exercise, you should carefully evaluate the tax consequences. By default, the IRS will consider you to have earned taxable income on the difference between the fair market value and the strike price as the stock vests. This can be disastrous if the stock does very well. However, there is an option (an “83b election” in IRS parlance) where you can choose to pre-pay all taxes based on the exercise up front. In this case the taxes are calculated immediately, and they are based on the difference between the fair market value and the strike price at the time of exercise. If, for example, you exercise immediately after the stock is granted, that difference is probably zero and, provided you file the paperwork properly, no tax is due until you sell some of the shares. Be warned that the IRS is unforgiving about this paperwork. You have 30 days from when you exercise your options to file the paperwork, and the IRS is very clear that no exceptions are granted under any circumstances.

I am a fan of early exercise programs, but be warned: doing early exercise and not making an 83b election can create a financial train wreck. If you do this and you are in tax debt for the rest of your life because of your company’s transient success, don’t come crying to me.

What if you leave? The company has the right, but not the obligation, to buy back unvested shares at the price you paid for them. This is fair; the unvested shares weren’t really “yours” until you completed enough service for them to vest, and you should be thankful for having the opportunity to exercise early and potentially pay less taxes.

Taxes

Taxes on stock options are complex. There are two different types of stock options, Incentive Stock Options (ISOs) and Non-Qualified Stock Options which are treated differently for stock purposes. There are three times taxes may be due (at vesting, at exercise, and at sale). This is compounded by early exercise and potential 83b election as I discussed above.

This section needs a disclaimer: I am not an attorney or a tax advisor. I will try to summarize the main points here but this is really an area where it pays to get professional advice that takes your specific situation into account. I will not be liable for more than what you paid for this advice, which is zero.

For the purposes of this discussion, I will assume that the options are granted at a strike price no lower than the fair market value and, per my discussion on early exercise, I’ll also assume that if you early exercise you made an 83b election so no taxes are due upon vesting and I can focus on taxes due on exercise and on sale. I’ll begin with NSOs.

NSO gains on exercise are taxed as ordinary income. For example, if you exercise options at a strike price of $10 per share and the stock is worth $50 per share at the time of exercise, you owe income taxes on $40 per share. When you sell the shares, you owe capital gains (short or long term depending on your holding period) on the difference between the value of the shares at exercise and when you sell them. Some people see a great benefit in exercising and holding to pay long term capital gains on a large portion of the appreciation. Be warned, many fortunes were lost doing this.

What can go wrong? Say you have 20,000 stock options at $5 per share in a stock which is now worth $100 per share. Congrats! But, in an attempt to minimize taxes, you exercise and hold. You wipe out your savings to write a check for $100,000 to exercise your options. Next April, you will have a tax bill for an extra $1.9 million in income; at today’s tax rates that will be $665,000 for the IRS, plus something for your state. Not to worry though; it’s February and the taxes aren’t due until next April; you can hold the stock for 14 months, sell in April in time to pay your taxes, and make capital gains on any additional appreciation. If the stock goes from $100 to $200 per share, you will make another $2 million and you’ll only owe $300,ooo in long term capital gains, versus $700,000 in income taxes. You’ve just saved $400,000 in taxes using your buy-and-hold approach.

But what if the stock goes to $20 per share? Well, in the next year you have a $1.6 million capital loss. You can offset $3,000 of that against your next years income tax and carry forward enough to keep doing that for quite a while – unless you plan to live more than 533 years, for the rest of your life. But how do you pay your tax bill? You owe $665,000 to the IRS and your stock is only worth $400,000. You’ve already drained your savings just to exercise the shares whose value is now less than the taxes you owe. Congratulations, your stock has now lost you $365,000 out of pocket which you don’t have, despite having appreciated 4x from your strike price.

How about ISOs? The situation is a little different, but danger still lurks. Unfortunately, ISOs can tempt you in to these types of situations if you’re not careful. In the best case, ISOs are tax free on exercise and taxed as capital gains on sale. However, that best case is very difficult to actually achieve. Why? Because while ISO exercise is free of ordinary income tax, the difference between the ISO strike price and value at exercise is treated as a “tax preference” and taxable under AMT. In real life, you will likely owe 28% on the difference between strike price and the value when you exercise. Further, any shares which you sell before you have reached 2 years from grant and 1 year from exercise are “disqualified” and treated as NSOs retroactively. The situation becomes more complex with limits option value for ISO treatment, AMT credits, and having one tax basis in the shares for AMT purposes and one for other purposes. This is definitely one on which to consult a tax advisor.

If you’d like to know if you have an ISO or NSO (sometimes also called NQSO), check your options grant paperwork, it should clearly state the type of option.

Illiquidity and being trapped by stock options

I’ll discuss one more situation: being trapped by illiquid stock options. Sometimes stock options can be “golden handcuffs”. In the case of liquid stock options (say, in a public company), in my opinion this is exactly as they are intended and a healthy dynamic: if you have a bunch of “in-the-money” options (where the strike price is lower than the current market price), you have strong incentive to stay. If you leave, you give up the opportunity to vest additional shares and make additional gains. But you get to keep your vested shares when you leave.

In the case of illiquid options (in successful private companies without a secondary market), you can be trapped in a more insidious way: the better the stock does, the bigger the tax bill associated with exercising your vested options. If you go back to the situation of the $5 per share options in the stock worth $100 per share, they cost $5 to exercise and another $33.25 per share in taxes. The hardest part is the more they’re worth and the more you’ve vested, the more trapped you are.

This is a relatively new effect which I believe is an unintended consequence of a combination of factors: the applicability of AMT to many “ordinary” taxpayers; the resulting difficulties associated with ISOs, leading more companies to grant NSOs (which are better for the company tax-wise); the combination of Sarbanes-Oxley and market volatility making the journey to IPO longer and creating a proliferation of illiquid high-value stock. While I am a believer in the wealthy paying their share, I don’t think tax laws should have perverse effects of effectively confiscating stock option gains by making them taxable before they’re liquid and I hope this gets fixed. Until then to adapt a phrase caveat faber.

Can the company take my vested shares if I quit

In general in VC funded companies the answer is “no”. Private equity funded companies often have very different option agreements; recently there was quite a bit of publicity about a Skype employee who quit and lost his vested shares. I am personally not a fan of that system, but you should be aware that it exists and make sure you understand which system you’re in. The theory behind reclaiming vested shares is that you are signing up for the mission of helping sell the company and make the owners a profit; if you leave before completing that mission, you are not entitled to stock gains. I think that may be sensible for a CEO or CFO, but I think a software engineer’s mission is to build great software, not to sell a company. I think confusing that is a very bad thing, and I don’t want software engineers to be trapped for that reason, so I greatly prefer the VC system.

I also think it is bad for innovation and Silicon Valley for there to be two systems in parallel with very different definitions of vesting, but that’s above my pay grade to fix.

What happens to my options if the company is bought or goes public?

In general, your vested options will be treated a lot like shares and you should expect them to carry forward in some useful way. Exactly how they carry forward will depend on the transaction. In the case of an acquisition, your entire employment (not just your unvested options) are a bit up in the are and where they land will depend on the terms of the transaction and whether the acquiring company wants to retain you.

In an IPO, nothing happens to your options (vested or unvested) per se, but the shares you can buy with them are now easier to sell. However there may be restrictions around the time of the IPO; one common restriction is a “lockup” period which requires you to wait 6-12 months after the IPO to sell. Details will vary.

In a cash acquisition, your vested shares are generally converted into cash at the acquisition price. Some of this cash may be escrowed in case of future liabilities and some may be in the form of an “earn-out” based on performance of the acquired unit, so you may not get all the cash up front. In the case of a stock acquisition, your shares will likely be converted into stock in the acquiring company at a conversion ratio agreed as part of the transaction but you should expect your options to be treated similarly to common shares.

On one hand, storage sounds like a great business: data volumes are growing exponentially with no end in sight, and as deployment architectures move to the cloud, centrally managed storage should become ever more important, right? Should this be a great decade of market growth for EMC and NetApp provided they can maintain their leadership in the sector.

I’m not so sure. I think there are three challenges that they face, of which two are under appreciated by the market.

The end of the database neutral-hardware vendor

First and most obviously, with Oracle’s acquisition of Sun, HP-Vertica, IBM-Netezza and of course EMC-Greenplum, the separation of data storage hardware and data management software has ended. It used to be that EMC and NetApp could ride in behind Oracle wins to sell some storage. Life has gotten harder.

Cost drivers in big data

Second, I think the big data wave will benefit storage vendors much less than people think. Stepping back from the hype, there are two reasons why people will eschew databases entirely in favor of something like Hadoop:

1. Complex analytics which require a custom program, not a database query
2. Large low-value datasets which aren’t worth putting in a database

I think there is a lot of value in the complex analytics, which should be captured by the Hadoop software ecosystem if they can avoid infighting and brutal price competition. I expect Hadoop will play a significant role for a long time for these types of applications.

The second Hadoop opportunity is also very large, though I expect it will monetize at a lower rate . In particular I don’t think people will cough up much money for higher performing storage behind these applications. And in my opinion, people are conflating the high-value case (complex algorithmic data mining) with the large-volume use case (a repository which is slightly smarter than just dumping things in files but much cheaper than a database).

Will people spend money for high-performance map reduce? Yes. And will lots of data sit in some sort of Hadoop (or Hadoop-like) systems? Yes, but most of it at very low cost and very low margin for storage providers. There is a good business in making Hadoop fast, its just not big enough to move the needle for a multi-billion dollar storage company.

Replicated local storage

Third, the database industry is changing. While replication has long been available and a reasonable way to provide greater reliability in the face of storage failures, it is becoming a more common approach. This is being driven by a combination of high cost for shared storage, high complexity in clustered filesystem based failover, and the general trend towards distributed databases for cloud deployments. I’ve seen a number of large deployments put on hold because of central storage costs and brought back to life based on distributed databases with redundant local storage. Not good for EMC or NetApp.

Can these challenges be overcome? Maybe. But its far from obvious to me that today’s high end storage vendors will benefit from big data and the cloud.

– Max

As technology executive with a broad range of functional experience, I am struck by the observation that most successful technology companies are run by their founders. Last year as I was considering running a company, it led me to ponder whether the move was futile: was I doomed by history to achieving a mediocre outcome by the simple fact of being a hired CEO? Happily, I found a situation at 10gen where I am President and partner with a founder/CEO that I like a lot; our roles are very complimentary, six months in things seem to be working well for both of us, and last but certainly not least the company is over-performing both of our expectations.

Tonight at a dinner event I found myself seated next to one of the handful of exceptions to my observation about the most successful companies having been run by founder CEOs. Naturally, I was eager to get his advice on the topic. His advice was simple:

• Remember its the founders’ company and you’re there to help
• Find a way to add value
• Build trust over time
• Don’t let your ego get in the way