SEO in a Nutshell:
- <2010: You could trick Google, and it was in your interest to do so.
- Today: You still can trick Google, but it’s becoming less in your interest to do so.
- >~2020: Forget it. No tricking Google. Be the “real-deal” or get out of the game.
Get Smart Or Get Out
Would you believe that Intelligent Robot Agents are on the way? Google Now, Apple Siri, Microsoft Cortana, Amazon Echo, Samsung’s S Voice and IBM’s Watson are all going to grow up; “into what?” is a much larger question. But in the meanwhile, expect things to become a lot more like talking to the Computer on the Bridge of the Starship Enterprise. Old-school crawl/index has already taken a backseat to all the other systems being “blending” into search in a first-page takeover (quick answers, knowledge graph, ads, PLA, etc, etc), and the transition from the old top-10 model to a completely and seemingly perfect auto-curated experience tailored to your history, the context of the question (location, web history, patterns, etc.) is just a matter of Google dialing-down the weighting of the older methods. And to trick the things doing the new-style autocurating is wayyyyyy harder than to trick the things doing the old-fashioned crawling & indexing; so don’t even try. Be prepared, be genuine, or be marginalized. Sorry about that Chief.
Machine Learning Optimization (MLO) vs. Intelligent Agent Bartering (IAB)
I’m going to take a crack at naming the next evolution of SEO. Probably, the industry will just justify making SEO continue to work to describe what we do, but I’m favoring MLO for Machine Learning Optimization, or IAB for Intelligent Agent Bartering. We’re dealing with intelligent agents like Google Now, Siri, Echo and Watson that will gradually always run-middleman in query transactions, and we will increasingly have to play to them, if we want auto-curation of the results to play out in your favor. We’re trying to play into machine learning algorithms rather than craw/index procedures. We need to think of Google as a jury of our peers, and our optimization efforts as amassing a preponderance of evidence, satisfying the more stringent burden of proof that should be necessary when trying to assemble the world’s best, certifiably relevant and accurate and happy-making results. Machine Learning Optimization… I like it. Go Google it.
Yep, doesn’t appear to be targeted by the likes of me yet. So, let’s go do some homesteading in the noosphere! M.L.O. Machine Learning Optimization. Search Engine Optimization (S.E.O.) is undergoing an undeniable transition. We all build our creations on the shoulders of giants, who in turn did the same. And so it goes, with our industry. My trick was spinning out .html files with xsl-document matches in XSLT transformations; I did a series-of-transforms that took long text files (raw bones) and slided and diced them up into delicately optimized interlinked and fully stylized HTML pages — all before CSS was even supported by mainstream browsers (i.e. I.E.). And so it was in the long-tail days, where more pages meant more precise targeting, and more hits. Alas, no more. Eric Schmidt proclaimed they were going to start cleaning up the cesspool of the Internet using brands as a key relevancy indicator, back in 2008 in front of a room full of magazine publishershttp://adage.com/article/media/google-s-schmidt-internet-cesspool-brands/131569/
And here we are seven years later, and we’ve seen (the oft-forgotten) Vince update that boosted brands, and the Panda and Penguin updates that fought spam, and the Caffeine infrastructure update that reduced reliance on map/reduce, and went to a more sort of real-time distribution of index-bits, while they got absorbed in the background, presumably, by more PageRanky traditional core-results updates. I remember how they used to talk about it with Inktomi, whose group I was the moderator for in the old JimWorld SearchEngineForums.com site, before his and its demise. Yup, I go way back. But the point stands, a big Net clean-up has occurred to the point where cost-of-hosting surpasses any AdSense money you can earn from having those pages published and out there. In time, spammers stop paying their hosting bills and their domain renewals, and… well, it’s sort of like a CB squelch dial being turned to reach silence with a sudden removal of all white-noise. There’s a threshold there, and an important lesson to be learned.
That lesson? Oh, Google’s gonna out your spamming ass, eventually. Googles gonna load you up with relevancy signals that spammers are going to have a difficult time faking, until all spammers are gone. The problem is Google not throwing out the baby with the bath water. How many sites out there are perfectly relevant and awesome, which will never move over to https protocol? Lots, I suspect. They’ll keep paying hosting, and they’ll keep getting traffic, oblivious to the changes in the industry going on around them, wiping out so many other peoples’ traffic that are not so privileged as to have a preponderance of evidence propping it up and giving it special get-into-search-results-free pass? So, the lesson is to BE THAT WEBSITE. Be the wizened old irreplaceable codger that’s been around forever, dispensing just the right wisdom on just the right topics within your domain space, to forever have bursty relevancy renewal overlapping, reinforcing, validating your past histories of was-also-relevant-then signals that Google has in it’s forever undo-history of the net, that would make The Internet Archive’s wayback machine look like a thumb-drive.
The time has come, the Walrus says to speak of may things. Of ships and sales and ceiling wax, of cabbages and machine learning optimizations… or MLO, for short. Sure, it also is my initials, but nobody ever reads too much into that, right GW-BASIC? Perhaps the strongest self-effacing testament to people naming software after themselves, after getting Linux out of his system, is his second act… git! Yep, Linus admits that git is named after himself. I’m sure you can Google up interviews where he jokes about this.
So maybe it is my own special optimization schemes. Maybe it does stand for me. Okay, then I better lay out some parameters of this proposed new field.
In your optimizations, you must first and foremost expect human spot-checking by the crowd-sourced remote-working quality assurance folks. Can’t fool humans.
Or can you? The skills of phishing have to come to marketing. It’s not enough to have a pretty, responsive site. You must be believable. Believability is key. Your web presence must put aside all doubt that you are anything but one of the preeminent voices to listen to in your space. You must exude authority, even if it’s unpolished and not slick at all.
There are many ways to do this. You should do many of them. It takes considerable work to broadcast out over multiple channels enough signals of relevancy-on-topics, getting secure certificates to prove you’re you and the authenticity of your message, and formatting to display nicely on phone, tablet or giant monitor… let’s see… what else? I’m not going to name all the signals. There’s plenty, though they’re ironically owned by different folks then the ones who need it most, and Facebook and Google will never share, and so there’s that permanent inefficiency in the digital marketing machinery that’ll never get efficient, least it invoke the wrath of anti-monopoly government breakup. Google and Facebook can only barely get away with what they’re doing now because the government doesn’t want to make too much noise about it, because they’re doing so much worse.
So, the field is named. It is indeed, Machine Learning Optimization, in the same way SEO was (is?) search engine optimization in its day. And so, I stake my claim here, sticking my flagpole into the noosphere, knowing in a moment I’ll git commit and push this puppy on up to Github, which’ll auto-publish it to the mikelevin.com domain, stylizing this one long texfile where I use markdown and a few HTML wrappers, so that strapdown.js kicks in and formats it as an (almost) readable web page. I do this all as just an experiment, where my commitment to the cause is undeniable, given Google’s ability to see and make inferences about the growth of this very file, which is just one of the many experiments I conduct.
Probe, probe, probe. Test, test, test. Infer, infer, infer. Have lots of irons in the fire, so that any one of them that starts to go your way can receive increased focus, increased thought-power going into precisely why and how that location is going to become better and better and better. Make those pages kick-ass! Do something innovative that’s only loosely related to web publishing, and make the published pages just a very well-optimized side-effect of the cool thing you’re doing. This becomes especially true, as the content within Apps themselves starts getting served in search, and the running of those apps achieved through search results, without even having to install the app (finally, some of the promise of Java achieved).
So many fronts on which to turn and face. So many channels and sub-channels of every type on every device at every size with every level of priority you can set, with every filtering-rule, with every special-alert. How is one to make sense of it all?
Well, it comes down to an active vs. passive thing. In sonar, there’s active and there’s passive. If you’re a sub trying to keep your location secret, you sure as hell aren’t going to send out a “ping”. That would advertise your location, like shouting “Hey, here I am!”. No, passive sonar is the thing… which is basically just sitting and listening. You can’t “light up” the space around you to echo-locate, but neither can the enemy see/hear you unless they bounce a similar ping off of you, advertising their location to you.
Everyone starts out in web marketing in passive sonar mode. You can hear everyone around you who is broadcasting their location to you — at least those that match your allow-criteria and don’t get filtered by your deny-criteria. You’re born just a consumer. But you have to learn to ping. You have to learn how to flip on that active sonar, and start directing it out in sometimes broad swaths (one channel), and sometimes narrow beams (another channel), and in beams of varying focus out to clusters or groups that you know to be there. It’s the new art-form of any-to-any broadcasting… narrowcasting… multicasting… casting by any other name, pours molten all the same.
Final thought before I wrap up this proposition to name the emerging field Machine Learning Optimization. Machine Learning is the new buzzword, as Search Engine was then. There are some competing terms that may be technically more accurate, but so too were there technically more accurate descriptions of what SEOs did than “optimizing search engines”. So, I’ll hold my ground and say that in precisely the same spirit, we are “optimizing machines learning”. Huzzah! And if that doesn’t convince you, then I fart in your general direction! Your mother was a hamster and your father smelt of elderberries! No, but seriously, have a sense of humor when you’re doing all this. It’ll go a long way.
One-Topic / One-URL… Forever!
So then, what is this page? It’s my beginning to deal with the new world. My SEO opinions are now to be on mikelev.in/seo and will now be forever. It only makes sense. And I will not subivide it into sub-pages, because why? So, welcome to the one-long-page I maintain on my site for SEO now that I’m going to maintain for life.
Why not? I mean, it’s my exact name plus exactly the topic–how lucky am I for the last two letters of my name to match up with a popular, easy-to-register top level domain (TLD)? It will go a long way to demonstrating the first most important principle of SEO: competition is fierce. The last thing you want to do is have to do all the work that goes into SEO for more than a handful of important URLs.
Consolidate, focus, and only try to do 3 or 4 things super-well… in fact, better than anyone else in the world. Keep the URLs short, memorable, and never-changing. Make the experience to be had on that URL world-class for the topic, and unlike any other to be found on the Internet.
It’s never too late to start. You can always rely on the world changing with very disruptive shifts brought about by Google and other forces giving you another chance to get your foot in the door, because tons of old competitors are going to fail to adapt to the disruptions. In fact, you can always rely on your competitors not adapting sufficiently to the new realities and position yourself to take advantage of that fact.
Start banking on principles and factors that you think will endure–like all the wonderful attributes of long-running, few-in-number URLs, sporting constantly evolving and best-int-the-world content. Cream rises, and impurities (spam) get filtered out.
This page is nowhere near what it needs to be, if I actually want myself to start to rise to the top of the search results on just SEO, but it’s a start. I say this now in 2015, which is already 20 years late to the game in the field of SEO, because SEO is older than Google. I was there at the time, optimizing for Webcrawler, AltaVista, and Infoseek (which could pick up changes almost instantly, even back then). All the names were different, and metacrawlers ruled the roost, blending the results from 3 or 4 engines for consensus, and briefly those sites were considered best. I was the moderator of the Inktomi forums on the JimWorld site, www.searchengineforums.com, the now defunct but still very first search engine forum site on the Web. Later, Brett Tabke’s Webmaster World took center stage, then later SEOMoz—now, just Moz.
Somewhere along the way, I came to New York to pursue SEO and made HitTail.com, still a healthy and valuable keyword tool today. It survived Panda and Penguin. It survived Google’s taking away of the keyword value in the http referrer variable. Then I went to work for 360i in the early-mid days, after their success in the SEM/SEO space, but right before their stellar rise in the online brand space and acquisition by Dentsu. While there, I helped get 360i a strong position in the 2012 Forrester Wave Report—now, feeling a bit dated. But fact remains, my proprietary technology that I build for them to crawl websites directly into Google Docs for real-time site analysis (among other functions) made 360i look favorable under the proprietary in-house built tools checklist column.
Yeah, so I got some SEO skills. I was on the team helping get JC Penney’s infamous penalty in Google lifted. I was the SEO Director on the Apple Online Store for some time. Kraft Food SEO was under my guiding hand for awhile. Before 360i, I worked for Hachette Filipacchi Media as an embedded in-house SEO for over a year on such properties as Car and Driver and Women’s Day. Before that was Connors Communications—the opportunity that brought me to New York, and where I made HitTail. Before that, I was in the multimedia software and digital signage industries with Scala Multimedia, a company forever-fewer people will remember from the Amiga days. And I worked for a tiny ERP-software company, called Prophet 21, now part of Epicor. I had diverse SEO experience, both in-house and agency-side, and sprinkled with invention along the way.
Constant Tectonic Shift
SEO is dead. SEO is not dead. SEO is dead again. SEO is just good business decisions and everyday marketing. Blah, blah, very zen if you build it, they will come Field of Dreams stuff. Gold rush! All obvious, valuable terms have been snatched-up but the early prospectors and the wise. Those that converted traffic into revenue, and those who didn’t sold or abandoned their old sites. The Web became a cesspool, says Erik Schmidt, then Google CEO, to a room full of magazine publishers in 2008, assuring them brands will rise again. Of course by then, everyone is thoroughly beaten up by Google’s disruption, and it is a mostly Google-game by the time the clean-up job is well underway (Panda, Penguin, Hummingbird, etc.).
This is always the case. This is nothing new. There is nothing spammers can do, no matter how clever, that Google cannot somehow algorithmically or crowd-source scrub away. And there’s no spam-fighting Google can do that can’t be tested at the edges and be loopholed-around by determined spammers. And so the arms-race of automation and counter-automation and the occasional clever leveraging of human labor (Q/A panels, captchas, etc). This is the role of art. Only a true artist, creating something that can somehow cut through the increasingly noisy (bad signal-to-noise ratio) Internet. Infinite channels. Infinite messages being blasted out over infinite channels. Finite number of humans. Finite amount of media humans can consume simultaneously.
Search is special. Search turns around the normal media proposition, which goes something like: I will sit here and passively consume you, and based on what you know about me from my watching habits and other online proclivities, I will insert advertisements into your experience, hoping to reach you with just the right message at just the right time. Push, push, push! Search is pull. Search is when a question pops into your head, and you turn to a machine to get an answer. The perfect timing of the answer-machine slipping in an ad at that very moment… well, that’s advertising value on a whole other order than mere push-ads. Bill Joy knew this with GoTo.com. Google now knows it with AdWords. I think pretty much everyone trying to compete with Google AdWords for ad-spending budget knows this. That’s why there’s so much desperation to breed a hyper-effective form of push, through programmatic.
But you get the picture? Search is special. Search isn’t going away. And search can’t be 100% commercialized into Google “product”—no matter how much everyone in the advertising industry would probably like that. In a twisted way, Google is fighting to maintain a sense of journalistic objectivity in keeping natural search results the algorithmic derivative product of a deliberately objective process. Originally, this was the Backrub algorithm, later relabeled PageRank. But PageRank was easily manipulated, and many more relevancy-ranking factors started becoming available as browser capabilities increased and link-sharing channels increased. The fight against spammers kicks in. The keyword portion of the referrer variable is removed from http requests, killing auto-optimizing SEO secret weapons around the world.
Constant tectonic shift. Nothing new. Disheartening sometimes. Pure SEO science leads to automated spam-cannons, eternally remixing scraped content. It’s a dystopian data form of grey ooze, and was the cesspool of which Erik spoke. So, SEO is less based on explicit evidence, and more based on a preponderance of implicit evidence. But correlation does not mean causation.
Google turns the squelch-knob, and some of my old SEO know-how goes obsolete. Don’t-be-like-Dad panic sets in. Where have I gone wrong? Have I relied to much on hackneyed tricks? Google now rewards far fewer landing-pages by default, unless you manage to overrule the quality safety-nets they’ve constructed and thrust your way into the still-there long-tail (hidden under magical incantations and verbatim-mode). But SEO must be concerned with the most common scenarios in search. Search-volumes must be chased, and one must still find ways to step into the path-ways of intent. This means data. This means much data from different sources, somehow knit together. This means quickly re-knitting data together in the formulaic ways that worked for you in the past… and constantly improving even that process.
Ugh!!! I need my secret weapons again.
But they need to be lightweight, immediately insanely useful, and somehow adaptable to whatever the future brings. I can not incur a major support liability. Build it. Distribute it. Nurture tribe around it. Parlay fame into fortune at some later date… even if it’s good street cred when job-prospecting.
I need to be active in Github. I need to be making myself the long-term authority in some niche, but still somehow delightfully valuable and enjoyable subject-matter.
It’s got to be something I can be passionate about.
Oh hey! Why not that lightweight Linux virtual machine you’ve been experimenting with, knitting together Tiny Core Linux with QEMU for a sort of Noah’s Ark Python execution platform. I’ll call it Levinux, and brand it as a Micro Linux for Education. Done.
But now I need a killer-app, and all 3 of my old tricks are plaid out:
A joyful agile framework to slam-out CRUD-apps… oh, Ruby on Rails has trashed any secret-weapon value my old Generalized System on VBScript may have had.
A system to slice-and-dice database-housed source data into artfully linked-together fully rendered HTML pages using an XML transformation system, powered by XSLT. Also done, but I’m no longer convinced of the wisdom of this method, when the same can now be easily accomplished with truly dynamic systems and caches caching cachable content. Good XSLT parsing that can output individual pages are really only in the Java world for some reason. And most content management systems are so much more optimized for search then they were ten years ago.
Strike, strike, strike!
Oh! APIs! I need an easy way to accumulate my API work under a unified user interface for rapid ad hoc work. I did something a little bit like this in the past. But I’ll do it again from scratch, and better this time, with a smaller software stack, and completely carried out in public under FOSS licenses on the Github site. I’ll make it the killer app to run on my Micro Linux.
I’ll call my Micro Linux, Levinux.
And I’ll call my API Killer Mash-up App, Pipulate.
And I will use them together to make myself an notable authority in several online neighborhoods, and have some secret weapons in my back pocket for SEO, and a grasp on technology that I can use to help me home-school my kid.
Constant tectonic shift can be fun if you’re adaptive. Or, would that be responsive?