10 ms·

> Instead of making a single big bet that might have focused the company on something completely different and potentially groundbreaking, Ms. Mayer staked out

> Instead of making a single big bet that might have focused the company on something completely different and potentially groundbreaking, Ms. Mayer staked out a lot of small and midsize positions, rarely committing to anything early enough to make a difference. For Ms. Mayer, original programming was just one of dozens of products in a portfolio that remains too complex to understand. Other sources indicate that Mayer is very much a metrics-based decision maker (a). I'm wondering if that's the result of putting a metrics-based decision maker at the head of a technology company. For sustaining innovation, carefully finding a local maximum seems fine, but in a dynamic field, you also need to be able to make a big move based on a more qualitative understanding of your surroundings. As a technical contributor, I've been on-and-off frustrated my whole career by quantitative managers asking for promises of real profit before committing to cleaning up technical debt. There's no honest way to come up with a good number here. You can't solve the problem thirty times and then present a high mean and low p-value. "Keep doing what we're doing but maybe 5% better" requires absolutely no proof, in contrast. You don't have to prove that COBOL-based system XYZ won't blow up in your face or that frames-in-frames webpage mycompany.com/abc won't eventually look unprofessional and affect the company's brand. I'm wondering if Mayer's "inability to bet the farm" is a higher-level version of the same problem... setting the bar too high for definitive proof creates barriers to strategic moves into areas with incomplete information. a) http://www.businessinsider.com/how-marissa-mayer-figured-out-work-at-home-yahoos-were-slacking-off-2013-3 http://www.businessinsider.com/how-marissa-mayer-figured-out...

iaw 11y ago
I think you have a valid, and insightful, point. Metrics are meaningless for visionary projects because there's nothing there to measure yet. I actually would have been happy if Yahoo resurrected a web portal and built social search to address the failings of Google, but instead there was this weird unfocused acquisition spree.
- 2close4comfort 11y ago
  I think you both are on to something. It does seem that there is a lack of focus and direction which I think makes and organization spread in any and all directions that they perceive profit. With focus in any one direction the org with have path on which they can build off of instead of running in all directions at once with no one ever making a real breakthrough. Being metric based just means that you are more focused on building shareholder value than making long-term decisions. That works well in some businesses, in others and I think this is more true in Technology this is not a valid long-term strategy.
  - ethbro 11y ago
    I think the gradient descent / machine learning metaphor is a good one, because it illustrate blind local optimization given a set of value functions. In other words: exactly the behavior that organizations over a certain size tend to exhibit (or put another way, the middle-managers-revert-to-the-consensus-value-approximation decision metric). As you note, the problem with tech is that for truly profitable endeavors, there is always a technical barrier that first much be overcome. And from first-glance, that seems like an impossible / overly-expensive burden (especially given the unknown reward behind it).
Touche 11y ago
Great post. Settings the bar too high is something I've seen many times. It's used as a way to give the illusion of collaboration and meritocracy if only you can prove something that is, in reality, out of reach.
jnbiche 11y ago
Yes, this is painfully true. Excellent comment. But is there a good way to overcome this, and convince cautious, quantitative managers that an decision with no corresponding data is truly necessary, and could yield great returns?
- davismwfl 11y ago
  In my experience, not generally. Cautious and quantitative managers usually became that way through careful management of every step, fearful almost of not being able to forecast and see the next steps. So by nature they are nearly impossible to convert. The one way I have seen it and participated in it, is when someone takes it in their own hands and does something bold to prove to the manager that there is traction and profit to be found outside the paint by numbers scheme. The problem, depending on the manager is that either the bold person is rewarded and the company moves towards some of that greener pasture, or the persons head is cut off and a good project dies right along with all employee morale. More simply to me this is the difference between an entrepreneur and a hired gun/manager/ceo. Most entrepreneur's are comfortable being uncomfortable, so they will make the "risky" call as people will see it, whereas a hired CEO may not because they are comfortable in measuring and proving to the board how good they are doing by coloring inside the lines.
Steko 11y ago
The article's conceit is that apparently she should have bought Netflix in 2012 which suggests some future article ideas for the author that practically write themself: maybe Wang Labs could have been turned around by buying out Microsoft in the 80's; maybe the railroad barons could have bought out Ford right as the model T was going into production, man we could do this all day! It's easy to be a visionary with 3 years of hindsight.
- ZanyProgrammer 11y ago
  And the possibility of such an acquisition ruining Netflix. I can't imagine a lot of Netflix engineers would've been happy to work for a dinosaur like Yahoo.
  - coldtea 11y ago
    Why, aren't they professional engineers, working for pay? Or do they go where its the most fashionable?
    - briandear 11y ago
      They are in demand. Why would you work for someone like Mayer (famously against remote working for instance) when they can have their pick from dozens of other companies that don't have a smug, self important failure as a CEO. For example, she had her child there in the office with her while denying other people the same ability. She decries working from home yet she simply builds a nursery there in her office. Could any other employee simply take over the next cubicle and install their nanny? Of course not. The grand irony is that remote friendly companies like Github, Basecamp and others aren't looking to sell off their core business. So sure Netflix engineers are professional, but they aren't stupid.
    - ergothus 11y ago
      This is a high demand field. I definitely work for pay, but I can get this pay at a number of places (To clarify: I don't work at Netflix or Yahoo) So I want additional things from my job like respect, involvement, empowerment, learning opportunities, flexibility. Generally those things are harder (though not impossible) to get in larger firms that are mired in communication weight. Speaking with zero first-hand experience with either company, Netflix probably struggles to handle those issues already, and to merge with a larger company that is trying to tighten its belt is unlikely to improve the situation. I've worked for 3 companies so far that merged with a larger, more "traditional" company, and every times there was high attrition from the engineering side as people saw various "good" parts of their job go away, from things as small as more bureaucratic procedures to as large as technology switches. I myself left one company when the new company required everyone to wear ties "because our sales people have to", despite the fact that we were in a separate site that no client would ever approach. That was 1999 and East Coast :) In retrospect, that seems a silly reason to leave a company, but it resulted in a significant pay boost and my period of greatest learning in my career. And I don't wear a tie. Now, both Netflix and Yahoo ARE tech companies, so my anecdotal evidence above may not even apply, but generally in this field demand for coders far outstrips supply, and anything that makes coders uncomfortable in large numbers can really threaten the status quo.
mseebach 11y ago
> quantitative managers asking for promises of real profit before committing to cleaning up technical debt I obviously can't speak for your experience in having these discussions, but I've found that the "debt" metaphor is extremely helpful in explaining this value to the business side (and, having worked in finance, it's actually more appropriately referred to as an unhedged call option which is much worse[1]). 1: http://higherorderlogic.com/2010/07/bad-code-isnt-technical-debt-its-an-unhedged-call-option/ http://higherorderlogic.com/2010/07/bad-code-isnt-technical-...
exelius 11y ago
> As a technical contributor, I've been on-and-off frustrated my whole career by quantitative managers asking for promises of real profit before committing to cleaning up technical debt. Agreed; though this is changing. More companies are starting to optimize their tech organizations for flexibility rather than low cost. The cost difference usually isn't that much (no more than 25%) but the focus is on building and maintaining a technical product development capability. All products today are technical products, and if non-technical companies don't develop a technical competency, they will wither and die (then their brands will be bought by whoever DOES have a technical competency). This is a strategic move, and strategic moves shouldn't be based purely off of numbers. They should be backed up by numbers, but market positioning and brand play into it a lot more than financial projections.
nickbauman 11y ago
In 2008 my 19-year-old nephew said "the future of television is in a website, not in the cable box". Even at the time this didn't even seem radical. In the end, ROI or metrics-based management approaches are mired in ideas that belong in businesses that resemble raw materials extraction (like, say, Exxon) or about as far away from a company like Apple or Google as you can imagine. When was the last time your company did an metrics improvement or ROI study on the phone system? The office electrical grid? Email? How the hell would you even conduct such a study? Would the iPhone project, prior to its release, survived such a study?
littletimmy 11y ago
Excellent post!
ksk 11y ago
>You don't have to prove that COBOL-based system XYZ won't blow up in your face or that frames-in-frames webpage mycompany.com/abc won't eventually look unprofessional and affect the company's brand. You can't tell someone they're doing it wrong if what they're doing is making them money. That is the bottom line. If my car is running fine, a mechanic telling me about my messed up gear alignment and replacing the clutch pressure plate thats 'about to die' isn't going to convince me to spend $3543 in parts and labor. That is what it sounds like to non-CS people. You can educate yourself about one aspect of car, but unless you're actually a mechanic yourself its trivial for an expert in a field to bilk someone outside of that field. As a developer, it would be super easy for me to make-up technical debt, talk about API complexity, brittle design and what not, but ultimately no manager is just going to allow a developer to go about changing the design for no perceived gain (to them). This is not necessarily a bad thing, as the vast majority of developers, are simply mediocre and are poor judges of what constitutes good design anyway.
- humanrebar 11y ago
  True enough. Your developers could be incompetent and lying to you, but they could just as easily be incompetent and/or telling you what you want to hear. > You can't tell someone they're doing it wrong if what they're doing is making them money. That is the bottom line. This is exactly the problem, though, isn't it? If it's easier to believe someone who tells you "we can keep muddling along indefinitely" than someone who tells you "this will cause problems eventually", then you have some sort of cognitive dissonance to counteract. In other words, what is the right response when your mechanic tells you "replace your breaks or you'll eventually die in a fiery car accident"? I'm pretty sure the answer isn't "well, I don't understand cars, so I'm going to defer this decision until you can quantify how soon I'll be dead".
  - ksk 11y ago
    >then you have some sort of cognitive dissonance to counteract. Well, you have to deal with the reality of how humans think. That cognitive dissonance is called the human condition. It is impossible to live life if one were to not have any biases or gut instinct. You'd have 1000 football games between two teams just to eliminate the different variables in order to determine which team was better. You'd have to visit a restaurant that served you shitty food, again and again till you made sure that it wasn't just a one-off, You'd have to allow a person to run a company into the ground multiple times just to make sure the previous failures were not the result of poor judgement, and so on. Chance and probablity rule our lives. Also, your brakes example is unsatisfactory. When people use brakes they get an immediate feedback which lets them know if they're working adequately. And even so, people are more likely to believe a mechanic when it comes to brakes precisely because they could die if they get it wrong.

iaw 11y ago

I think you have a valid, and insightful, point. Metrics are meaningless for visionary projects because there's nothing there to measure yet. I actually would have been happy if Yahoo resurrected a web portal and built social search to address the failings of Google, but instead there was this weird unfocused acquisition spree.

2close4comfort 11y ago
I think you both are on to something. It does seem that there is a lack of focus and direction which I think makes and organization spread in any and all directions that they perceive profit. With focus in any one direction the org with have path on which they can build off of instead of running in all directions at once with no one ever making a real breakthrough. Being metric based just means that you are more focused on building shareholder value than making long-term decisions. That works well in some businesses, in others and I think this is more true in Technology this is not a valid long-term strategy.
- ethbro 11y ago
  I think the gradient descent / machine learning metaphor is a good one, because it illustrate blind local optimization given a set of value functions. In other words: exactly the behavior that organizations over a certain size tend to exhibit (or put another way, the middle-managers-revert-to-the-consensus-value-approximation decision metric). As you note, the problem with tech is that for truly profitable endeavors, there is always a technical barrier that first much be overcome. And from first-glance, that seems like an impossible / overly-expensive burden (especially given the unknown reward behind it).

Touche 11y ago

Great post. Settings the bar too high is something I've seen many times. It's used as a way to give the illusion of collaboration and meritocracy if only you can prove something that is, in reality, out of reach.

jnbiche 11y ago

Yes, this is painfully true. Excellent comment. But is there a good way to overcome this, and convince cautious, quantitative managers that an decision with no corresponding data is truly necessary, and could yield great returns?

davismwfl 11y ago
In my experience, not generally. Cautious and quantitative managers usually became that way through careful management of every step, fearful almost of not being able to forecast and see the next steps. So by nature they are nearly impossible to convert. The one way I have seen it and participated in it, is when someone takes it in their own hands and does something bold to prove to the manager that there is traction and profit to be found outside the paint by numbers scheme. The problem, depending on the manager is that either the bold person is rewarded and the company moves towards some of that greener pasture, or the persons head is cut off and a good project dies right along with all employee morale. More simply to me this is the difference between an entrepreneur and a hired gun/manager/ceo. Most entrepreneur's are comfortable being uncomfortable, so they will make the "risky" call as people will see it, whereas a hired CEO may not because they are comfortable in measuring and proving to the board how good they are doing by coloring inside the lines.

Steko 11y ago

The article's conceit is that apparently she should have bought Netflix in 2012 which suggests some future article ideas for the author that practically write themself: maybe Wang Labs could have been turned around by buying out Microsoft in the 80's; maybe the railroad barons could have bought out Ford right as the model T was going into production, man we could do this all day! It's easy to be a visionary with 3 years of hindsight.

ZanyProgrammer 11y ago
And the possibility of such an acquisition ruining Netflix. I can't imagine a lot of Netflix engineers would've been happy to work for a dinosaur like Yahoo.
- coldtea 11y ago
  Why, aren't they professional engineers, working for pay? Or do they go where its the most fashionable?
  - briandear 11y ago
    They are in demand. Why would you work for someone like Mayer (famously against remote working for instance) when they can have their pick from dozens of other companies that don't have a smug, self important failure as a CEO. For example, she had her child there in the office with her while denying other people the same ability. She decries working from home yet she simply builds a nursery there in her office. Could any other employee simply take over the next cubicle and install their nanny? Of course not. The grand irony is that remote friendly companies like Github, Basecamp and others aren't looking to sell off their core business. So sure Netflix engineers are professional, but they aren't stupid.
  - ergothus 11y ago
    This is a high demand field. I definitely work for pay, but I can get this pay at a number of places (To clarify: I don't work at Netflix or Yahoo) So I want additional things from my job like respect, involvement, empowerment, learning opportunities, flexibility. Generally those things are harder (though not impossible) to get in larger firms that are mired in communication weight. Speaking with zero first-hand experience with either company, Netflix probably struggles to handle those issues already, and to merge with a larger company that is trying to tighten its belt is unlikely to improve the situation. I've worked for 3 companies so far that merged with a larger, more "traditional" company, and every times there was high attrition from the engineering side as people saw various "good" parts of their job go away, from things as small as more bureaucratic procedures to as large as technology switches. I myself left one company when the new company required everyone to wear ties "because our sales people have to", despite the fact that we were in a separate site that no client would ever approach. That was 1999 and East Coast :) In retrospect, that seems a silly reason to leave a company, but it resulted in a significant pay boost and my period of greatest learning in my career. And I don't wear a tie. Now, both Netflix and Yahoo ARE tech companies, so my anecdotal evidence above may not even apply, but generally in this field demand for coders far outstrips supply, and anything that makes coders uncomfortable in large numbers can really threaten the status quo.

mseebach 11y ago

> quantitative managers asking for promises of real profit before committing to cleaning up technical debt I obviously can't speak for your experience in having these discussions, but I've found that the "debt" metaphor is extremely helpful in explaining this value to the business side (and, having worked in finance, it's actually more appropriately referred to as an unhedged call option which is much worse[1]). 1: http://higherorderlogic.com/2010/07/bad-code-isnt-technical-debt-its-an-unhedged-call-option/ http://higherorderlogic.com/2010/07/bad-code-isnt-technical-...

exelius 11y ago

> As a technical contributor, I've been on-and-off frustrated my whole career by quantitative managers asking for promises of real profit before committing to cleaning up technical debt. Agreed; though this is changing. More companies are starting to optimize their tech organizations for flexibility rather than low cost. The cost difference usually isn't that much (no more than 25%) but the focus is on building and maintaining a technical product development capability. All products today are technical products, and if non-technical companies don't develop a technical competency, they will wither and die (then their brands will be bought by whoever DOES have a technical competency). This is a strategic move, and strategic moves shouldn't be based purely off of numbers. They should be backed up by numbers, but market positioning and brand play into it a lot more than financial projections.

nickbauman 11y ago

In 2008 my 19-year-old nephew said "the future of television is in a website, not in the cable box". Even at the time this didn't even seem radical. In the end, ROI or metrics-based management approaches are mired in ideas that belong in businesses that resemble raw materials extraction (like, say, Exxon) or about as far away from a company like Apple or Google as you can imagine. When was the last time your company did an metrics improvement or ROI study on the phone system? The office electrical grid? Email? How the hell would you even conduct such a study? Would the iPhone project, prior to its release, survived such a study?

littletimmy 11y ago

Excellent post!

ksk 11y ago

>You don't have to prove that COBOL-based system XYZ won't blow up in your face or that frames-in-frames webpage mycompany.com/abc won't eventually look unprofessional and affect the company's brand. You can't tell someone they're doing it wrong if what they're doing is making them money. That is the bottom line. If my car is running fine, a mechanic telling me about my messed up gear alignment and replacing the clutch pressure plate thats 'about to die' isn't going to convince me to spend $3543 in parts and labor. That is what it sounds like to non-CS people. You can educate yourself about one aspect of car, but unless you're actually a mechanic yourself its trivial for an expert in a field to bilk someone outside of that field. As a developer, it would be super easy for me to make-up technical debt, talk about API complexity, brittle design and what not, but ultimately no manager is just going to allow a developer to go about changing the design for no perceived gain (to them). This is not necessarily a bad thing, as the vast majority of developers, are simply mediocre and are poor judges of what constitutes good design anyway.

humanrebar 11y ago
True enough. Your developers could be incompetent and lying to you, but they could just as easily be incompetent and/or telling you what you want to hear. > You can't tell someone they're doing it wrong if what they're doing is making them money. That is the bottom line. This is exactly the problem, though, isn't it? If it's easier to believe someone who tells you "we can keep muddling along indefinitely" than someone who tells you "this will cause problems eventually", then you have some sort of cognitive dissonance to counteract. In other words, what is the right response when your mechanic tells you "replace your breaks or you'll eventually die in a fiery car accident"? I'm pretty sure the answer isn't "well, I don't understand cars, so I'm going to defer this decision until you can quantify how soon I'll be dead".
- ksk 11y ago
  >then you have some sort of cognitive dissonance to counteract. Well, you have to deal with the reality of how humans think. That cognitive dissonance is called the human condition. It is impossible to live life if one were to not have any biases or gut instinct. You'd have 1000 football games between two teams just to eliminate the different variables in order to determine which team was better. You'd have to visit a restaurant that served you shitty food, again and again till you made sure that it wasn't just a one-off, You'd have to allow a person to run a company into the ground multiple times just to make sure the previous failures were not the result of poor judgement, and so on. Chance and probablity rule our lives. Also, your brakes example is unsatisfactory. When people use brakes they get an immediate feedback which lets them know if they're working adequately. And even so, people are more likely to believe a mechanic when it comes to brakes precisely because they could die if they get it wrong.