The long view

On what time hori­zon do you work? With 20 years of the web now behind us, our field has a fair amount of his­tory, which has left us with a few the­o­ret­i­cal bar­na­cles to shake loose. But it also raises ques­tions as to the time hori­zon over which pre­dic­tions ought to be made and roadmaps designed. And it raises ques­tions on what our indus­try obsesses about in the here and now as opposed to what it could think about in the long run.

His­tory will teach us… something?

When Tim Berners-Lee first demon­strated the tech­nolo­gies behind what came to become the world wide web, he had lit­tle idea of the impact his inven­tion would have. But what has hap­pened since may allow us to look back and bet­ter under­stand how to look for­ward. While it took a cou­ple of years for the web to move from aca­d­e­mic and research cir­cles to a more main­stream and com­mer­cial­ized space, some of the early bat­tles are still reflected in what we see today.

For exam­ple, there is a con­stant strug­gle for con­trol of the view­ing expe­ri­ence on the inter­net. In the ear­lier days of the web, that strug­gle was rep­re­sented by the browser wars with a large fight explod­ing between the dom­i­nant browser cre­ator at the time (Netscape) and the dom­i­nant Oper­at­ing Sys­tem ven­dor at the time (Microsoft). The same fight can be seen today in the strug­gle over whether web appli­ca­tions or ded­i­cated plat­form spe­cific appli­ca­tions ought to rule the roost in the future.

The death of Netscape left the web bar­ren for a few years as few con­tender to the supremacy of Inter­net Explorer emerged until the Fire­fox project finally found its foot­ing. Over those years, the web largely stag­nated, partly as the result of a defla­tion of the finan­cial bub­ble that had arisen over the pre­vi­ous year but also partly as a result of the lack of a cred­i­ble con­tender to the dom­i­nant browser brand.

Look­ing for­ward by look­ing back

When look­ing at the future, we can be a lot more hope­ful today, because the odds of a winner-take-all have been low­ered in some seg­ments of the mar­ket. For exam­ple, the rise of iOS has been coun­ter­bal­anced by the rise of Android and an increase in the num­ber of web-based apps. With such a diver­sity in the mobile space, it also will make it dif­fi­cult for new entrants to suc­ceed. Mar­ket­place tend to reward only a cou­ple of groups and we are already at a point where we have 3 play­ers. This is bad news for Microsoft and Nokia who are try­ing to play this from behind and have a real uphill fight to gain any seri­ous trac­tion, no mat­ter how great their offer­ing is.

In the tablet space, the story is a lit­tle more wor­ri­some. Apple’s cur­rent dom­i­na­tion of that mar­ket could lead to a new stag­nant space unless a con­tender is iden­ti­fied. Amazon’s posi­tion­ing of the Kin­dle Fire as a poten­tial alter­na­tive may present oppor­tu­ni­ties to restore a bal­ance, forc­ing all play­ers to bring their best game. At this point, it appears that the mar­ket­place has widely rejected other offer­ings in the tablet space so Amazon’s entry makes this an inter­est­ing case. The ques­tion at this point is not whether Ama­zon will top­ple Apple (it won’t) but whether it will intro­duce a prod­uct that appeals to a seg­ment of the con­sumer mar­ket and forces Apple to keep improv­ing its offer­ing (and here, I would present the iPod roadmap as an exam­ple of how inno­va­tion dies in a mar­ket­place, even if that mar­ket­place is dom­i­nated by Apple: from a prod­uct stand­point, the iPods sold today have not really changed since the intro­duc­tion of the iPod Touch four years ago).

On the stan­dard front, we are see­ing some fierce bat­tle lin­ing up in how browser sup­ports new fea­tures of HTML5 and related tech­nolo­gies. As I pointed a cou­ple of weeks ago, this is a good thing. The cur­rent diver­sity of browsers in the mar­ket­place forces every player to attempt efforts at being on the cut­ting edge. This is the kind of thing that will speed up adop­tion of new tech­nolo­gies across the land­scape as a whole and will help the open web become a strong con­tender for the future of com­put­ing. Tech­nolo­gies like HTML5, CSS3, WebAu­dio, and WebGL are help­ing the web become more oper­at­ing system-like and, as such, increase its chances at becom­ing a strong player on any platform.

Roadmap 2025

But look­ing at today and look­ing for­ward, we are see­ing a mar­ket­place that is increas­ingly frag­mented. So what’s a startup with lim­ited resources to do? Should one put all their eggs in a sin­gle bas­ket and bet on a sin­gle plat­form? Should a com­pany look at intro­duc­ing fewer fea­tures across all plat­forms, spread­ing their prod­uct roadmap over a longer time­line? Or should devel­op­ers go to the low­est com­mon denom­i­na­tor and tar­get through the web browser?

The answer is it’s com­pli­cated but, at its core, one needs to think about what is needed to move the com­pany for­ward and get it to reach early mea­sure­ment mark­ers while still focus­ing on the future.

At Keep­skor, we call that Roadmap 2025.

I know it sounds extremely ambi­tious for a com­pany to think to 2025 when it doesn’t yet have a prod­uct in the mar­ket­place. But the issue for most startup is not ambi­tion but the lack of such thing. By focus­ing on Roadmap 2025, we get a sense of how the mar­ket­place is evolv­ing and how we want our prod­uct to evolve accord­ingly. Fre­quent read­ers of TNL.net know that this col­umn is not really con­cerned about the next few months and tends to focus more on the under­ly­ing cul­tural and busi­ness trends that can have decade-long impacts. This is why things like the impact of DVRs on the tele­vi­sion busi­ness were dis­cussed here over a decade ago, or more recently, I looked at what chal­lenges existed for books and tele­vi­sion. The 2025 con­sumer mar­ket will be impacted by those changes and the way in which appli­ca­tions designed on the inter­net inter­act with that will also be impacted. As will the dom­i­nant plat­forms, one of which, we hope, will be Keepskor.

But how does a Roadmap 2025 look in the present? Well, put quite sim­ply, the present is an early alpha. The min­i­mum viable prod­uct the mar­ket­place will accept for some­thing much larger. But the MVP for a plat­form that will exist in 2025 is sub­stan­tially big­ger than the MVP for a sim­ple app. And build­ing it requires higher lev­els of abstrac­tion than a nor­mal prod­uct. For exam­ple, no one really knows what the win­ning front-ends will be in 2025. What we know is that they will be dif­fer­ent than the ones we are used to today. So we design our plat­form to ensure that input and out­puts are sep­a­rated from every­thing else. In the short run, this cre­ates an added layer of trans­la­tion that increases latency (mea­sured in mil­lisec­onds) to every trans­ac­tions but a slow­down we con­sider accept­able for now.

Iter­ate backwards

The basic agile method of devel­op­ment looks to scrums that are mov­ing the prod­uct for­ward in an iter­ated fash­ion, tack­ling one com­po­nent or another over short cycles. The level of clar­ity as to what will hap­pen in the next scrums is gen­er­ally depen­dent on how far for­ward you’re look­ing as each scrum can move a prod­uct to the adja­cent pos­si­ble. In a long roadmap model, one iter­ates back­ward. For exam­ple, if you are look­ing at what your prod­uct looks like in 2025, do you know what it looks like in 2024? You may have a rough idea as to how the prod­uct should evolve along the way. Maybe you get a sense of what it looks like in 2020 or 2015?

I call this back­wards iter­a­tion. Move from the ideal state and start plant­ing base camps back­wards. It’s a strat­egy I learned from long-distance skat­ing (I’ve done runs of up to 65 miles on a sin­gle day). No one wins a long dis­tance race by focus­ing on the end point. The win is only by focus­ing on the next 100 yards and iter­at­ing those for­ward. But before you do so, you need to have a sense of what you need to do by each of those 100 yard mark­ers. And that means con­sid­er­ing how tired you will be and how much slower you will be at the end of the race and work­ing back­wards to assess the best bal­ance of expand­ing and deplet­ing energy vs. yards won.

In a busi­ness, this allows you to build a model that is pretty accu­rate for the first cou­ple of years, and only mildly less so after­wards. It also allows you to go back to the roadmap when major events hap­pen and iden­tify whether they will have any impact on your busi­ness (and if they do, whether this impact needs to be mit­i­gated now, and can be mit­i­gated by mov­ing an item for­ward in the roadmap).

From a com­mu­ni­ca­tion stand­point, it also makes it eas­ier for peo­ple get­ting involved in your busi­ness to get a sense of where you’re look­ing to go. This may mean that some investors are not inter­ested because they don’t buy into your vision (and that’s OK because those investors would be an issue fur­ther down the road if you’re still going down some of the same paths) and also means that some peo­ple will not be inter­ested in join­ing the com­pany because they don’t agree with the steps mov­ing for­ward. I con­sider that early fil­ter­ing as it ensures that your com­pany gets proper buy-in from all stake­hold­ers, thus low­er­ing the chances that employ­ees will leave once they are fully committed.

Study the past, under­stand the future

But a roadmap 2025 plan is not some­thing that one pulls com­pletely out of thin air. What it is, more than any­thing, is a set of assump­tions based on close study of part events. For exam­ple, the major­ity of the Keep­skor roadmap can be explained by events that have already hap­pened. This ground­ing in his­tory and his­tor­i­cal trends helps us ground the busi­ness into mod­els that have been proven to work in the past and which, when com­bined, turn out to be sub­stan­tially big­ger than the ini­tial ideas.

When look­ing more closely at his­tory, I get a sense that most areas of his­tory have some cycli­cal com­po­nent to them, with a pen­du­lum swing­ing in one direc­tion or the other but always mov­ing us for­ward ulti­mately. The study of pre­vi­ous cycles is prob­a­bly the great­est tool one has in under­stand­ing what the future can look like.

How­ever, there is one thing to note: The dis­tant future is eas­ier to prog­nos­ti­cate than the closer one because things actu­ally do change at a much slower pace than most would expect. In the 1990s many dot­com­mers, present com­pany included, made what looked like wild state­ments about the future and how the inter­net would change every­thing. Circa 1999 and cer­tainly in 2001, those seemed like crazy pro­nounce­ments but, in hind­sight, the world we described as com­ing to bear at the turn of the cen­tury is pretty much the world we live in today: the rev­o­lu­tion has just taken longer to take hold.

A per­sonal note

For years, peo­ple have asked me why I was writ­ing TNL.net as I didn’t seem to derive any real income from it and the things I wrote about were, on the whole, only tan­gen­tially related to what I do. Over the last year, I’ve come to dis­cover that the writ­ing of TNL.net is a some­what self­ish act. First, it allows me to crys­tal­lize a lot of my own think­ing: the process of writ­ing and edit­ing an entry forces me to focus on a topic area very closely. It also forces me to review source mate­r­ial and his­tor­i­cal data to ground and inform my think­ing. And then, there’s the part where you, the reader, enter. Your emails or dis­cus­sion help me get a bet­ter sense of some of the flaw in my argument.

Update: I gen­er­ally start writ­ing early in the week and fill the entry up over time until it gets pub­lished on Sun­day evening. This week, Fred Wil­son posted “Long Roadmaps”, a great entry that cov­ers some of the same ground. I guess there must be some­thing in the air here in NYC that gets us to think about those things.