Rethink Mobile Search, 10–100x faster. Introducing WonderSwipe 👆🏼

Search Different.

Gary Fung
12 min readJan 20, 2018


UPDATE: WonderSwipe has launched on the iOS App Store since 4/18! Get it now for free. Read on for why I think it is the biggest thing to search since Google itself, and watch our 30s demo video below for how. If you think it’s awesome too, tell us and vote for us here.

WonderSwipe’s icon, and demo video

There’s too much stuff online & too little time on the go to fuss with browser tabs or going back & forth in browser history.

When was the last time you did some serious googling on your smartphone? According to stats like this and the domination of time spent on mobile over desktop, not likely to be recent or frequently. For me, not unless I’m stuck with mobile Safari or Chrome without a laptop. Mobile websites, including Google itself, stuffing more ads above the fold on smartphones’ limited screen space compared to desktop haven’t helped the UX.

👉🏼 So, here’s 3 scenarios to use WonderSwipe instead of conventional browsers, so mobile search stop sucking:

  1. Too slow; Didn’t search?
    10–100x article loading and browsing speed, as measured in-app. Even more drastic difference over slow data networks.
  2. Too many results; Didn’t browse?
    No tabs: Easily swipe through search results, no clicks or background tab tab tab tab tabs needed.
    Hands-free mode: Automated browsing, with search result summaries read out like audiobooks and scrolled in-sync.
  3. Too long; Didn’t read (TL;DR)?
    Reader mode (like Pocket, Safari’s) with summarization of each search result, where possible.

After 2 years of development, I believe my solution is ready. WonderSwipe is designed from the ground up to find the answers you search for, in the fastest way possible. With the Internet’s enabling of prolific information overload and the recent backlash against Facebook (newsfeed), Youtube (autoplay), Snapchat (snapstreaks), etc. We want to create the opposite of a time sink, for truly time well spent. We want to create new, efficient and meaningful ways of information searching, curating and sharing. We are a counterpoint to time spent maximalists like the ad-driven social apps, we want to be time minimizing. Here’s demo videos of WonderSwipe’s phase 1 launching on the App Store, which I’d say is about half feature complete.

WonderSwipe demo (in HD, on mobile), on iPhone X — searching 🔍 dunkirk date before or after pearl harbor
iPhone 6 version

Note that WonderSwipe is not a browser. It’s something different, a research and reference tool. It augments search best when your search is non-trivial, and turns up informational articles to which WonderSwipe can do text extraction and summarization. WonderSwipe’s phase 2 will add highlighting and annotation for your own reference, as well as making it sharable in a social layer so learners can help other learners like you, quickly sifting and surfacing diamonds in the search results rough. For every topic searchable.

This web scale social highlighting we believe will shift us towards more meaningful sharing and time well spent in learning and curating, by going beyond simplistic likes and hearts that festers in viral engagement metrics, driving sensationalism, fake news and time sinks for advertising.

WonderSwipe for now at launch is the not-browser, with the sole goal of making search and finding your answers buried in results ridiculously fast. It does include in-app browser (SafariView on iOS) as a 1-tap fallback when you want the full HTML for some websites to function. (globe icon on bottom right of app, in the demo video)

My hope in developing this after isoHunt, is that it be useful to people to use more than once a week. By that metric, after intense internal testing (myself 😅) for the last few months, I can say that this has been my first project to pass that bar. I found myself using WonderSwipe for any non-trivial search more useful than mobile browsers. And when I want to have result summaries read out to me, I’d use the app even when I’m on my desktop.

With my 11 years experience working on the BitTorrent search engine isoHunt, and before that, using and watching Google’s development since its beginning. I believe WonderSwipe is shaping up to be the biggest user experience improvement in Web search, since Google itself. And that begins with us waking up to the fact that mobile search right now, is sucking.

More and more computing time shifts from desktop to mobile, while time spent on the Web in current browsers shrink.
Mobile web browsing haven’t fundamentally changed much. And all serious web browsing start with a search.

The Web’s Problem

With the proliferation of content, web search is getting more tedious. It’s exacerbated on smartphones. I would tap on a promising search result, wait 5 seconds (rather optimistic on the best cell network conditions), read a bit, see that it doesn’t concern what I’m looking for, tap back in browser to Google’s search page, scroll to another promising search result, rinse and repeat. It’s just too tedious and slow to do any serious search on a non-trivial topic of interest.

This pattern is similar on desktop browsers, though it’s better with the aid of faster WiFi, easy switching between browser tabs (so you can background load search results without all the click and wait), browser extensions, and easy Cmd-F / Ctrl-F to highlight keywords within found webpages for quick skimming.

But what if you can conduct mobile search 10–100x faster, with all the niceties I listed above in desktop browsers?

Well all except one, WonderSwipe can’t give you fast WiFi everywhere. But it does compensate for slow networks, possibly going beyond desktop access speed in the overall search experience. (SX for short from here on, as in UX)

WonderSwipe’s Solution

This is what WonderSwipe offers, for your knowledge presenting super-smart phone:

  • Navigate from search result to search result, by swiping through them horizontally, like you would swipe through a gallery of photos. The cornerstone of WonderSwipe is this completely gestural 2D UI. (more below)
  • Right after Google return a page of search results, WonderSwipe goes to work, loading all 10 results in the background, in parallel. So the first result you swipe to may still take 5s like normal browsers (but likely less), further results you swipe to will take no wait time.
    This is the worst case of 10x faster knowledge access I stated in title.
  • Each result will either be presented with the original HTML and Javascript of the webpage (above worst case for load time), or if detected to be an article with a main body of text to extract (powered by the former Readability), that text will be further summarized and presented instead.
    This article text extraction + summary avoids Javascript, HTML rendering and early image loading, which cuts result access time by another factor of 2 on average, even with the added mobile CPU processing in summarization. Blocking of ads and otherwise undesirables in articles that require Javascript is a side-effect, and necessary for speeding up access for users like you, without compromise.
  • If article summaries are presented, the full text of the webpage can be viewed by tapping one of the summaries. The paragraph containing that summary would be scrolled to in the origianl text.
  • If an article was found by your previous search, or a search by any other user of WonderSwipe, the result with summary is cached in our cloud. Results from our cloud cache would take only ~0.5s to load! That’s another factor of 10 in reduced load time, for a total of up to 100x faster knowledge access.
WonderSwipe’s debug console, logging timestamps of the ideal case. Here, 9 cached results load in parallel within a total of ~0.5s, in total.

And, bonus:

  • Keywords of your search query are highlighted on each search result, in HTML webpages, extracted article texts or summaries.
  • WonderSwipe stays in portrait mode while you read, even if you’re fussing with toilet paper and what not on your quest for knowledge and inevitably and unwittingly rotate your phone. You could read search results expanded in a full Web view, and only there is landscape mode enabled for images and videos.
  • Hands-free mode. Turn your search results into an audiobook! On any summarized search result, tapping the 📢 speaker icon turns on text-to-speech mode. The current summary sentence in view is read, and the summaries scroll down as they are read. When all summaries of a search result are read, WonderSwipe “swipe” itself to the next summarized search result, read out the article title, and proceed reading its summary. This continues until you tap the speaker icon again to stop, or it reaches the end of up to 10 results on the current Google search page.
  • Thumbnail images of search results are preversed and showcased as bouncy headers. On summarized articles, other found images are listed together below the summaries.
  • Extract web links in copied text from another app, and do a 1 tap search for that link, summarized and all. (🔍 Paste: <link> on bottom)

A note that my up to 100x faster knowledge access benchmark is network dependent (obviously). Due to WonderSwipe’s approach of article text-first-and-only loading, in parallel, further accelerated by cached summaries, the overall improvement may actually be even more profound when comparing WonderSwipe vs. conventional browsers over particularly slow, intermittent mobile networks.

I’ve also not quantitatively included further knowledge transfer improvements, resulting from article summaries and search keyword highlights as those are somewhat subjective. WonderSwipe’s current naive summarization algorithm compresses word count to ~40% of original text on average, so consider at face value another 2x factor in speeding up article reading. And depending on your speed reading ability, keyword highlights is a great aid in speeding up text scanning further still, to extract meaning as you skim over search results. So I could bump my benchmark up to 200x or more, but who’s counting what’s already an astronomical improvement? 😅


And what about voice search, you say? Browsing search results is so last millennium. Yes that’s useful, but that only speeds up the query input part in a serious topical search. If it’s a trivial search like this, WonderSwipe isn’t the solution it’s designed for. You’d find the answer directly on the first page of search results, using WonderSwipe or any old-school browser. Or you could indeed just ask Siri or Google or Alexa or Cortana.

If your search is non-trivial however, like “is AI a threat to humanity”, that’ll take some digging beyond even the first page of search results, and voice assistants are unlikely to give a simple answer (or any answer). Siri for one defaults to web search when it can’t give a simple, factual answer, and that happens often. You may not find any good answer while googling either, and you’ll have to refine your search query. All of which involves skimming search results to decipher relevance, and this kind of non-trivial search is exactly what WonderSwipe is designed for.

Design Philosophy

WonderSwipe came out of my realization that mobile search is slow. Too slow, and we oftentimes don’t bother until a later time on a desktop, or none at all when you forget. What’s the point of the Web being a Great Library of human knowledge, if our computing paradigm moved from desktop to mobile, and we take a collective step backward and no longer bother accessing this library?

The problem is speed of access and knowledge transfer.

There is more to the Web than timber.

On the other hand, WonderSwipe is not a browser replacement. It’s a superset, focused around search. Not all search results are articles with text that can be extracted and summarized. When WonderSwipe fails, on web apps for example that cannot function without Javascript, the original HTML webpage is rendered by default, or you can load it explicitly by opening Web view within the app.

WonderSwipe attempts to extract meaning (article text) wherever it can, and when it does, it further apply algorithmic summarization to solve your TL;DR problem, which is especially apparent on the go. So that’s a further cognitive load reduction, beyond the measurable 100x load time reduction in the entire Search Experience.

In optimizing SX, I realized the workflow of the standard browser is one dimensional. You click a link, you go back, you click another link, you go back again or you click on a new link. It’s a linear timeline which you go back and forth in. But why be limited to a 1D UI, when you can swipe in 2D and flow from one search result directly to the next? Lists of search results are already conveniently sequenced for this, WonderSwipe just lays them all out horizontally. Swipe left or right for the next or previous search results, up and down to get into the summary and images or full text of a particular search result. Perfect, fluid and completely gestural in 2D, as it’s meant for the modern smartphone.

Which brings us to the name of the app. In making swipes the singular centre of user interaction without buttons, it becomes a wonder to access the knowledge of the Web. And infinitely so — on reaching the end of search results 1–10, another swipe left automatically searches the next page on Google. Further swipes continue to results 11–20, 21–30, etc.

And in each search result, for articles that split themselves into multiple webpages that you’d normally have to tap “Next” in-between, WonderSwipe also attempts to fetch all article pages and present them as 1 singular body of text, which gets summarized together. Nifty!

Attention is the New Scarcity

To optimize for SX completely and without compromise, I suggest breaking the current loser proposition of the user in the Web ecosystem. For you the user are not Google’s customer, nor are you for most web publishers. If you didn’t pay them, you are just eyeballs for their true customers, the advertisers. Your convenience in reading and optimizing for knowledge access isn’t their priority, distracting you with ads actually is.

But your attention is valuable.

This is an attention deficit crowd.
credit: cloudflare

More valuable than ever, with the Web’s overflow of information. Attention is the new scarcity, as we get less and less bored. Thus, optimizing SX from the ground up in the user’s perspective is all the more important, and demands fundamental rethinking. The extraneous gunk websites carry, either intentionally for monetization, or unintentionally with non-optimal heavy design or otherwise legacy design that’s not responsive to the mobile small screen and slower networks. Without fixing them holistically client-side, the mobile web continues slipping into irrelevance. There’s a saying: “The Capitalists will sell us the rope with which we will hang them.” Let’s stop buying their ropes.

This means optimizing knowledge extraction on the Web to its logical end. By dropping Javascript altogether — ads, tracking and interactivity, WonderSwipe is able to achieve that 10–100x access speed-up without compromise. Web view is a button tap away for search results that need it, but serves only as a fallback on your explicit request.

Which brings us to the mission statement… 🥁🥁🥁

The Future

To make WonderSwipe an even bigger wonder, with our mission to Make the Web Yours. Here’s my rough roadmap: (I’ll write more about them later)

  • Improve article text extraction and summarization. I’m currently just using untuned, off the shelves libraries.
  • Offline referencing and search, on locally cached prior search results.
  • A social curation layer for every web search: Bookmarking, highlighting and annotation. Privately useful first, communally beneficial second — a private research and note taking tool, while also reduce friction from search query to knowledge for the whole Web.
  • Q&A, for the truly hard search queries with no relevant results found. Think lifelines like phone a friend on Who Wants to Be a Millionaire. Yes there’s Quora, but I’m looking for a more open system, and one more integratable with search queries.
  • Explicit sharing that’s way more useful than LMGTFY. Think embeddable, curated search results.
  • Personalized/recommended trending searches. Top new interesting searches of the day for some communal discovery.
  • Extensions to supercharge certain categories of websites with additional, useful data. Example: price history on shopping search results, fake news check. An open API for evolving a new open standard.
  • Build/use a better general search engine than Google (maybe). Bing, DuckDuckGo? Add other specialty search engines like Amazon for different verticals (definitely).
  • Adopt BATs, if it doesn’t fizzle out.
  • Create a KnowledgeCoin to let users fund authors, curators directly. If the ICO bubble hasn’t burst first.
  • Resize app for (Apple) watches, tablets, TVs, AR glasses, VR googles.
WonderSwipe on a demo search (30s video)

Questions and comments? I want to hear them! I can’t do this without your input. This is the most exciting idea and project I’ve started since isoHunt in 2003. I hope you find WonderSwipe as useful as I do.

Read more about our thinking about problems we want to solve in search and knowledge sharing, by following us here on Medium (top of page), or:

You want to just download WonderSwipe and give it a spin? For those who signed up for beta, you should have received an invite email with code to redeem WonderSwipe on TestFlight. If not, WonderSwipe has launched on the iOS App Store on April 18.

We are working on our Android version now. Stay tuned. I’ll leave you with a choice quote, which says much with few words the problem of our time:

A wealth of information creates a poverty of attention. ~Herbert A. Simon



Gary Fung

Founded isoHunt, WonderSwipe. Web3 building @ and 70% engineering 30% product