Hard is not defensible (2017) | Hacker News Reader

Hard is not defensible (2017)

alexcrompton.com

79 pointssaadalem52 comments

> In startups, being defensibile means competitors can’t catch you once you’re ahead.

I'm not sure this is a useful concept. Investors are burning billions of dollars trying to ensure that Uber-or-Lyft [or insert other unicorn battle here] gets un-catch-ably "ahead", meanwhile neither is a successful business in the traditional sense. There are tons of successful businesses in the world that just execute well and deliver value in exchange for money. They're not succeeding because they are "ahead" of the competition, they are succeeding because there is plenty of room for multiple competitors to succeed.

The difference is whether or not that business has network effects that reduce the viability of competitors.

For example, in the Uber-or-Lyft 'battle', sure, there are and have always been private car companies and private drivers, but when you want a car you're likely to open one or maybe two apps on your phone and call the quickest/cheapest option. Similarly, Facebook competitors have a hard time getting off the ground because when you're the only one in your friend group on a new social network, there's no one to talk to.

Meanwhile, a new email provider has a small amount of work to do to catch up with all the security regulations, but then can just provide value to its customers without having to worry about 'defensibility'. A single app aggregating Uber, Lyft, taxi, and 3rd-party car services on one screen would open those up for competition. Similarly, a federated Facebook that provided various social networks in one client would let you pick your hosting provider, configure your own feed, and communicate with people on other platforms.

Sure, but I think it's stupid to be so focused on creating the next network-effect-monopolizer. There just aren't as many opportunites to do that as VC money seems to think. In the Uber-or-Lyft example, most drivers drive for both. Nice network monopoly you've got there.

> A single app aggregating Uber, Lyft, taxi, and 3rd-party car services on one screen would open those up for competition. Similarly, a federated Facebook that provided various social networks in one client would let you pick your hosting provider, configure your own feed, and communicate with people on other platforms.

Is there any talk of requiring companies that have or intend to have effective monopolies due to network effects to make their services price-comparable (or aggregatable, in the case of the likes of Facebook where the "price" is your being served ads) in some way as you mentioned? It seems like it would do a lot to solve the problem. Not being able to shop around is antithetical to competition

Uber’s problem is people living in NYC etc mostly just stay in NYC or other major city. The app advantage is only really meaningful for travelers which make a small slice of the overall market. A low cost local competitor in NYC would rapidly get New Yorker’s to swap over and it could then grow one city at a time.

For now everyone is just waiting until Uber and Lyft run out of money, but longer term it’s really not looking good for them.

I thing this article is missing a third type of hard: time-consuming/long-tail hard. Google solved a hard problem, but it's a technical problem and it could be replicated by another company, particularly since there are probably a lot of developers who have experience working on Google's Search. But the problem is that there is an absurdly long tail of features (curated news sources, relevant social media results, answer boxes, etc) and being able to serve that long-tail is part of what makes Google so valuable. It would take years for even a well-funded team to catch up to Google in this regard and the high barrier-to-entry is absolutely a moat.

I think that's just a type of bodybuilder hard. It just takes a lot of effort, discipline and time to do it.

I disagree. Bodybuilder hard at described in the article is just a matter of effort. High barrier-to-entry means that you need extreme resources to start competing. I think they are quite different.

There is also social-hard in the network effects of incumbents and familiarity to overcome.

Having resources is just a different thing, they’re not a kind of work. Getting the resources is work, sure, but it takes either cleverness (maths hard) or effort (bodybuilder hard) to get them, or is not hard at all because you already have them. But actually building the things you cited is bodybuilder hard.

That's really not true. You can try as hard as you want but you will not be able to spin up a search competitor right now because the funding literally doesn't exist.

Even suggesting you want to tackle Google's lead in search or ads as a startup will get you laughed out of the room.

The amount of time, effort, and funding needed is well beyond what any investor wants to even consider.

There are search competitors in non-US markets. The lack of funding begs the question of whether Google's search lead is simply due to no one trying to compete within the last 10 years.

DDG doesn't really qualify here as they aggregate google/bing search results.

How about Bing? Yes I know, they're easy to forget about, but....

Bing started 15+ years ago as a feature for feature clone of Google. Historically they've simply crawled Google results and replicated whatever Google returns. It could be a data point that Google has made the perfect search system as of 15 years ago - or just a footnote of the balmer era of Microsoft like Windows phone.

DDDG has raised $13 million. Neeva has raised $37.5 million. Hulbee has raised $9 million.

Resources aren't a kind of work you can do, they're not an activity.

I think DDG and Bing do all of that as well and in many cases better with low budgets.. The moat is that content creators do whatever Google wants and debug experience problems only coming from Google since it has the overwhelming number of users.

DDG definitely does none of those things. And yeah, Bing has funding and has been in development for over 11 years (and has a basis that goes even further back than that). But people don't really use it and that is partially due to the horrible reputation it developed when it came out. Bing is literally a joke and in many ways that is due to their struggles catching up to Google.

I think this only works out for google due to their business model and scale. Like if Google was a subscription service, would people pay 5x more for the long tail? I'm guessing not. I think the reason that the long tail is viable at google is that the higher value searches happen infrequently. You buy cars, houses, insurance, less often than furniture or toothbrushes and the ads are priced accordingly. So in essence you need a problem where the value and the frequency are inversely related.

For an even more extreme example, think of something like Wolfram|Alpha.

Any sufficiently hard problem is indistinguishable from impossible.

There is also the fact that there’s only a finite number of pages you can crawl a day. So Google has a built-in data/time advantage. They started from 1999 and have 2 decades of data already. Anyone starting now needs to first build up that huge index and that is almost impossible if you start now.

I agree with the article, but I also disagree with investor focus on defensibility over hardness.

If you are investing in a generic software startup at an early stage, it is much more reasonable to ask them:

"How many months/years of head start do you have over the competition, and how will you use my money to solidify this head start into an unfair advantage in your market?"

than to ask them:

"What stops someone else from doing this?"

Because if they are an early stage, generic software startup, chances are that the honest, no bullshit answer to the second question is, "Nothing." And that doesn't mean that it would be a bad business to invest in.

I fully agree, but I've yet to meet a VC that really appreciates honest, no bullshit answers.

I've talked with plenty VCs and my impression is they strongly prefer founders that are happy to bullshit them, and ask questions particularly to test that.

My theory is that the good ones do this on purpose, and the bad ones got bullshitted so often that they got used to it.

I don't think it's a terribly bad strategy even. If you're looking for Travis Kalanick style "growth at all costs" type founders, then brutal honesty is pretty low on the list of important character traits.

> I fully agree, but I've yet to meet a VC that really appreciates honest, no bullshit answers.

That's because you need to be able to effectively bullshit your customers and users. If you can bullshit the VC, well then surely, you can bullshit anyone.

Bodybuilding hard seems to conflate 2 types of hard tasks: move-a-mountain-with-a-shovel hard, where something is time consuming, such as having to hand-label millions of images for machine learning datasets, and quit-heroin hard, where something is against human desires, such as convincing people to abandon their current way of doing something for something else that is a short-term loss but a long-term gain, for example switching to a more performant but incompatible CPU architecture.

Math hard, as the article says, is not really defensible, since copying is easy. Move-a-mountain hard is more defensible along the time/money axes, competitors need to throw time/money at the problem to catch up. Quit-heroin hard is the most defensible, since customers will loath to pay the switching price regularly.

Of course, it's not as black-and-white as this. All products have a mix of the three, and its where the product lands in the mix that determines the promotion strategies that the company should take.

I like how you identified the hard not to do factor

Couldn't Google be an example of this? Search is a hard problem, even if you've got MSFT-level resources. There are definitely economies of scale and some weak network effects that help companies like Google, but I'd argue it's most because they do something hard better than anyone else.

Search is a hard problem, which increases the barriers to entry. But once a competitor has a method to crawl the Internet, that particular defensibility is gone.

I'd argue the data network effect is what gives them most of their defense. Now that they have a corner on what creates better search results, this is harder for others to defend against.

They're not dominant because it's hard though. Bing is similarly capable, but even if it gained a clear edge would it really win out over Google? Market dominance has little to do with hardness. Making a new cola isn't hard, but brand recognition and ubiquity are huge advantages on a different axis.

At least in Coke’s case, defensibility comes from the fact that making cola is hard due to regulation and supply chains. Coke is grandfathered into being allowed to import coca leaves (they now remove the cocaine) and buys up alot of the worlds vanilla supply. There is a reason even Pepsi can't replicate the flavor. You still have a great point about the value of Coke’s branding / marketing.

Habit is also massively important in what products /brands people use. Even if there is a superior, cheaper product many people will continue using the same product because that's what they always done.

One product that seems bodybuilding-hard is Google maps. When Apple maps had its calamitous introduction, I remember claims (by Google itself, I think), that Google was a decade ahead, and would remain ahead for the foreseeable future.

But for several years now, it seems to me that Apple maps has been more or less on par technically, and the case for rolling their own maps instead of relying on a third party has, if anything, gotten stronger.

All you have to do to be "defensible" in many industries is observe the status quo, but add decent customer service.

>> It’s unrealistic to think that you’ll ever have a totally unique insight, let alone one other people couldn’t come up with if they really wanted to.

These gentlemen would beg to disagree:

https://en.wikipedia.org/wiki/Wright_brothers

Before the Wrights, numerous others had tried and failed to achieve what they achieved. Of course, once the Wrights succeeded (or rather, some time after) their idea was free for the taking. But the Wrights did have a "unique insight" that was only their own.

Tangentially-addressed in the PS at the bottom:

> PS. My colleague Zefi Hennessy Holland raised an insightful point about hard to find market secrets that I can’t phrase better myself.

> He said ‘If Math Hard is finding the solution to a problem many people know about, Market Hard means finding an industry problem that few people know about through a lot of hard work… Your competition can copy your moves but not the underlying insights that made you make them.’

Uh... free for the taking is not exactly right. They patented the shit out of it. Look up the "Wright brothers patent war" they waged world wide for licensing fees. It took WW1, with the USA extremely far behind in aircraft innovation, for the gov to end their patent trolling adventure.

The article says its unrealistic; not that its impossible.

Its unrealistic to tell every star High School athlete that they'll become the next million-dollar NFL player. But, its not impossible; in fact, they bring on million-dollar players every year. The point isn't that its impossible; its that its improbable.

I would tend to think the key is the "soft" aspects of your business. The things that can't simply be copied or commoditized. Customer service, being in tune with your target market and their wants and needs, responsiveness to feedback, institutional knowledge.

Everyone still uses Slack even though it is not an especially hard product to build, and even though Microsoft and Google have both shipped clones that work fine and integrate with their other products. Why? Part of it is brand recognition and a network effect, but I think an important factor is that this is all that Slack does. Slack really really cares about making the best chat tool for people to use at work. They put all of their energy into constantly improving it; they give it their human attention. The same is partly true of Dropbox vs Google Drive and OneDrive.

Google Meet and Microsoft Teams don't really matter to those companies. They'd like it if you used them, they'll allocate some cash to keep them going. But it wouldn't be a huge deal if they lost the battle and those products died. A company's culture, institutional knowledge, and focus can't simply be copied.

I highly recommend the book 7 Powers by Hamilton Helmer. Helmer defines "Power" as the ability to retain meaningful profit margins for a meaningful period of time. Each of the 7 Powers is a method he has seem companies use to retain this (network affects, switching costs, etc.). Very insightful read.

Tbh, it looks like they are ignoring the elephant in the room: intellectual property...

I don't think patents are defensible against FAANGs either.

Yeah, this article is assuming a competitive environment when in reality there exists an arsenal of state tools available for corporations to entrench themselves as monopolies: IP protections, subsidies/contracts, regulatory capture, etc etc.

I think there are a few issues with this article.

Firstly, body-building hard ignores the fact that a competitor can't build up muscle overnight, and every day they spending building their strength is a day you continue building on yours. They need to be able to catch you, which they can't do.

Another analogy would be a marathon run. If you're already half-way through the marathon when others are just starting, the competitors need to have a strategy or capability which lets them run faster than you.

It's another way to look at why you need to be 10x better than a competitor. You need to be moving faster than they are.

"Math-hard" is not defensible in the US because the patent system has been weakened so much. VCs in the 1980s and 1990s looked for a strong patent position. VCs today look for a strong marketing position. VCs as a class made money in the 1980s and 1990s made money. That stopped working around 2000, when throwing money at something to buy market share took over.

(Yes, this will be modded down by the "IP should be free" crowd.)

One thing that isn't mentioned is that the 'intrinsic' body-building-hard isn't fixed for everyone.

Obligatory "Beating the Averages" http://www.paulgraham.com/avg.html

Can somebody explain to me why we care about these classifications, and why I wouldn't want to divide these into more (slightly) arbitrary groups?

What about Intel?

Silicon is bodybuilding-hard. It is extremely hard so there are few competitors, but it is inherently a series of extremely expensive coordination problems. To churn new architectures every couple years takes an oiled machine of thousands.

Don't forget that Intel had a few decades where nearly all software was written for their x86 chips. This reinforced every computer needing to have intel, and only recently has the move to web apps made it easier for devices to use something other than intel.

This is dead on. I've also heard this referred to as "tactics" and "strategy". "Tactics" are various bullet-lists of things you need to repeatedly do to maintain best practices (eating protein & lifting). "Strategy" is generally longer term and requires clever thinking (like Pythagoras).

Reply on news.ycombinator.com