I've found myself in many conversations over the past few years comparing etexts and printed texts (hereinafter 'ptexts'). While I have known and loved a great many ptexts, I am a fan of etexts, on the whole, with some exceptions (for now). The problem with such comparisons, as in so many discussions of technologies, is that there is enormous diversity within each modality and, at least in the case of etexts, the goalposts are moving all the time. ptexts are highly evolved and have reached a state of refinement that etexts will take a while to catch up with, though they are, in principle, manifestly superior in almost every way. In practice, there are plenty of etext technologies that are pretty awful and that give the medium a bad name, and plenty of scope for improvement in all of them. This is a common problem with disruptive technologies: near the start of their lifespans they tend to be worse than what they replace in at least some respects, diversify in all sorts of ways that are not sustainable, and often inherit skeuomorphic patterns of thinking and assumptions from their predecessors. PDF, for instance, is reminiscent of horseless carriages in the early days of automobiles, inheriting technologies like pages and layout techniques from ptexts that are mostly pointless and silly, though occasionally useful where layout and design matters, and offering new capabilities as well as duplicating old ones. DRM is a similarly clunky attempt to make a new technology behave like an old one and, interestingly and disturbingly, has spawned new and nasty frills like limits on period of availability or who can own an etext as well as replicating old patterns. As technologies evolve, they tend to fill in the gaps and, eventually, often improve on their predecessors. However, some things will likely remain lost forever when a new technology largely replaces an older one. The codex (the pbook format that we are most familiar with), for example, though offering great benefits in compactness, indexing and random access, has never caught up with the scroll that it largely superseded in at least one respect: variable medium width. A fixed page size was and remains a limitation of a codex that the refinement of folded concertina pages only partly and clunkily addresses but that a scroll (and an etext) manages effortlessly. 

With that in mind, here are a few thoughts on areas where and how etexts need to catch up, as well as where they already exceed the capabilities of ptexts.

How ptexts are currently better (but are losing ground all the time)

• context: it is easier to remember reading things in ptexts and sometimes to find them again, in part because of the physical contextual cues about where you are in the text (not just the thickness of the number of pages before and after but also subtler cues like the bending of page that results from that) , in part because of consistent formatting (it is much easier remember the shapes of text on pages, a feature that is lost when text is reflowed) and in part because of micro-cues, like imperfections on the paper or how the ink is applied. Search in an e-text requires a layer of abstraction beyond that visceral and immediate recollection - you have to know the words you are looking for, not the shapes. There is scope for improvement here. Some ereaders provide better context-sensitivity than others, often skeuomorphically, but it would not be a bad idea at all to research ideas like one I suggested a while back, such as overlaying subtle consistent background patterns or imperfections over the text to give a better sense of context.
• smell: books tend to smell very good, especially old ones. The smell of a library or used bookstore is hard to beat. On the other hand, I am similarly fond of the smell of old electronics. I love the smell of old valve radios, for example, or the waft of burning dust emitted by an old CRT. I suspect this is an association rather than anything to do with innate pleasantness. Likely currently emerging generations will wax lyrical over the smell of a new iPhone.
• power source: ptexts are powered by the sun or artificial lighting. Mains power is too limiting while batteries are a bane of modern life and progress towards better ones is frustratingly slow. Batteries run out all too quickly, catch on fire, damage the environment and get less efficient with age, though researchers have found plenty of alternatives involving capacitors, graphene, wireless power transmission, solar power and a host of other technologies that promise much and that have been shown to work for years or even decades. Even using current mainstream technology, would it really be so hard to put a solar panel on the back of every battery-powered smartphone or tablet so that it lasted a little longer? Like a month rather than a day or less? Or capacitors that could charge in seconds rather than hours? All of this is possible now, with little compromise apart from a little increased cost and size that I for one would be more than happy to accept, given long-term savings. I suspect a systemic failure here - batteries that die are a business opportunity for everyone from smart-device manufacturers to battery makers and, even if that were not an issue, there is massive momentum and drag from the infrastructure that supports the industry.
• size: some things we read need to be big, and current etext technologies tend to be limited in size, or are too heavy and power-hungry if not. This is especially true of graphics, tables, and charts, that tend to be less useful in the current generation of etexts. I look forward to improvements here in both standards and devices. Roll-up screens, smartphone projectors, smart glasses etc are all possible if a little costly now, though each has weaknesses so far. When very high resolution devices of any size are competitively priced, I suspect the solution is not a single device but to simply have multiple devices, some of which are huge. A thin, light, cheap, 4K tablet of (say) broadsheet size, lasting for months on a single charge, would make a lot of sense and is not so very far off. 
• moisture resistance: most (not all) ptexts last better in the bath than most (not all) smart devices. It is not expensive or difficult to waterproof most e-reading devices. I have plastic containers designed for this for most of my devices, but there are more elegant waterproofing options for many devices and some are already at least water resistant out of the box. It is already possible to read an etext under water pretty easily, while that will likely never be true of ptexts. ptexts also get wrinkly and mouldy if exposed to water vapour for too long.
• dust resistance: see water resistance. However, dust and sand can be the devil for electronic devices in ways water can only aspire to, while causing little trouble for ptexts. This will improve as time goes by, and can be overcome (at a cost) already, but is likely to remain a problem for quite a while. 
• device obsolescence: ptexts are pretty self-contained and remain readable for as long as they last. Encoding and storage standards for etexts change a lot. This is both good and bad: as a digital file, etexts can theoretically last forever and be immortal, but it requires active maintenance for that to happen, including transfer to different devices as well as occasional changes to encoding formats. Energy, in literal and management terms, needs to be put into it.  
• DRM: this is a completely unnecessary and deeply ugly invention that hampers some etexts but affects almost no ptexts. It should be removed and banned forever. Serves no useful purpose and hampers the spread of knowledge for no good reason. There are tools to remove it for most formats but not all, but cynical badly considered legislation in several countries may still stand in the way. Related to this is the hugely irritating habit that some publishers/distributors have of limiting how much can be copied from an etext, or keeping notes in a proprietary system, or limiting the reading tools that can be used to a single application. This is just wrong. 
• self-containment: ptexts need nothing more than light, reading skill and eyes to read them. etexts need a device. That's unlikely to change much and, if it did, we might lose the benefit of immortality of etexts.
• compatibility: related to self-containment, there are many standards for encoding etexts, and not all are trivial to transfer from one device or application to the next (though Calibre does a very good job of handling most).  Some evil developers deliberately use proprietary formats to prevent this. I do not like having my reading list fragmented by format and wish all such developers that do not change their ways to suffer bankruptcy in the worst possible way. 
• reading in direct sunlight: visibility in sunlight is not a problem at all if you have an etext reader with an e-paper display and not a huge problem with most modern tablets and phones, but heat and electronics are still not a great combination. This will change and is getting better all the time, but older generations of e-readers and computers suffer greatly in the sunshine.
• non-glowing screens: it is not good for sleeping patterns to read screens that glow with cold, blue-ish light, and the jury is out on whether it is bad for your eyes: concentrating on any fixed light source for a long time is probably a bad idea, but tablets, phones and e-readers tend not to be fixed. Of course, not all e-text readers are like this. Some, such as those using e-paper, do not glow at all, or let you turn the glowing off. For those that are stuck with the glow, tools like f.lux provide a brilliant way to reduce this problem and arguably deal with it better than the alternative of reading a book by artificial light. 
• distributed cognition: my bookshelf constantly reminds me of what I have previously read, sparking associations and connections. I can very easily organize and reorganize it in ways that help reinforce and augment those associations. It is a constant reminder of things that I know and ideas that inspired me. Etexts in a folder or an app are nothing like as evocative and require me to actively seek them out. I look forward to epaper wallpaper that will allow improvements over ptexts that are orders of magnitude better. Sometimes I dream of a touch-sensitive colour epaper wall that I can configure with any digital artifacts (photos, notes, books, etc, etc) in any way that I choose - a bookshelf, gallery, whiteboard and many other things, constantly updateable and ever changing. I could even configure it to read a book. This is technically feasible today but the costs would be rather prohibitive. A projection system might be a weak substitute but it would be power-hungry and would glow: glowing is not always a good thing in one's immediate environment. E-paper wallpaper is definitely the way to go. 
• rivalry: giving or lending a ptext seems generally more meaningful and poignant than sharing a digital artifact, at least partly because it is a rival good, while an etext is (in principle) non-rival. I can and do give ebooks as birthday gifts, but it feels less personal than a pbook with an inscription on the inside cover. This can have social and aesthetic benefits too - a coffee-table book, for example, becomes a social object that both communicates something about you and acts as a centre of conversation when people visit. ptexts are also pretty good for things like title deeds and things requiring signatures, though digital signatures work pretty well too and can be duplicated, shared and stored more reliably and easily.
• startup cost: though vanishing fast, there are still economic divisions in the world that mean some people cannot afford even the cheapest ereader. This is not going to last long. Prices are dropping fast, I have already seen them for less than $30, and even cheaper ones are available in some parts of the world. Moreover, within a couple of years there are likely to be more smartphones capable of ereading than there are people in the world (in fact, there already are more mobile internet devices than people). But there are features that a cheap reader will lack, at least at the moment, and a single pbook remains cheaper than an e-reader. Of course, once you have purchased a handful of pbooks you will have paid more than someone who purchased an ereader, who will also immediately have gained access to millions of ebooks for free. Swings and roundabouts here. If I had $30 to invest in improving access to knowledge for someone living in poverty in a poor locale, I think it would be better spent on an e-reader than a book. But if all I had were a dollar, an old book might make more sense.
• reliability: on the whole, ptexts crash less often than devices used to read etexts. I doubt that this will change that much for the foreseeable future. On the other hand, a crashed (burnt, water-damaged, ripped, etc) ptext is dead forever, while an etext can rise from the ashes every time. It can be inconvenient in the short term, though, if the only device available to you crashes, runs out of batteries or breaks, and that is far more likely to happen than the destruction of a ptext.

How etexts are already much better (and getting better all the time)

• device flexibility: I can read the same etext on many devices and it is often effortlessly portable between them, keeping track of what I was last reading and the notes I have made. As a geek researcher and teacher working in the field of learning technologies, I already have more than 20 devices on which I can read the same books, from a 3-inch phone through to a 40 inch TV, and most sizes in between, as well as many screen types, resolutions, brightness, portability, operating systems, physical designs etc. A ptext has one and only one format and it is often far from convenient. Of course, not all devices work well for etext: a laptop or a desktop computer, for example, is an atrocious device for reading etexts. 
• presentation flexibility: depending on the device and application I am using, I can read text of any size, kerning, line spacing, margins, font, colour, background colour that I choose. I love that I can adjust this to my eyes rather than vice versa. I also love that most e-reading apps can read to me using text-to-speech, though not always as beautifully as I might wish. Of course, not all apps work as well as one another for this, and there are still some display technologies that need a lot of work. 
• accessibility: etexts are vastly more accessible to people with different abilities and disabilities than ptexts. There are devices that read aloud, display braille, assist those with learning disabilities, assist those with motor disabilities (Stephen Hawking can use one), adapt to those with dyslexia, offer large type for those with poor eyesight, and plenty more. 
• intertextuality: hyperlinks are a good thing, great for following references and tracking counter-arguments and inaccuracies. Equally, being able to look up the meaning of any word on the fly or explore it further on the web (if available) is brilliant. For books used for learning, etexts are transformative, providing different ways of engaging with text.
• multimedia: we are not limited to images and text any more. Video, animation, simulation and interactivity are simple to incorporate. Not always a good thing, but nice to have the option.
• search: what is lost in contextual information is gained in text search. Who needs a page reference when you can look up any words or phrases in the text?
• updates: as new editions or corrections appear, they are available to replace or augment the text on our devices. Good publishers (which also tend to be those that eschew DRM, as it happens) give lifetime updates, as well as accompanying materials and resources.
• annotation: no more ugly squiggles and ragged bookmarks and sticky notes, though the equivalents are still available - you can just choose whether or not to show them. And the notes are typically available across multiple devices, and can be extracted, processed, aggregated, searched, integrated and shared with others. DRM can rear its ugly head here though - it is possible but difficult to get such things out of Kindle, for example, and not all e-readers support freehand drawing. Even those that do may have less value due to the ability to resize and reflow text that takes away the original context.
• repurposing: it is much easier to customize, re-use and repurpose an etext than a ptext, though DRM can rear its ugly head to try to prevent it. etexts are fluid, if we want them to be, and can be remixed in manifold ways.
• cost: ptexts are far more expensive the etexts, once you own a device capable of reading etexts. This relates not just in terms of the high environmental and financial cost of getting paper in the first place, but also the infrastructure needed to support them, from delivery trucks to bookshelves and the not inconsiderable space taken up in my limited accommodation by highly flammable and fragile paper. etexts are cheaper to produce and cheaper to store and, apart from where unjustifiable extortionate pricing practices are involved, cheaper to buy. For anything out of copyright, they are virtually free, give or take a little network, processing and storage use. Yes, devices cost money and don't last forever, and there are energy costs and media management costs that differ from those needed for ptexts, but devices are typically multi-functional. If any one of those functions beyond reading text is worth having, etext comes for (almost) free.
• durability: ptexts decay, get lost, become illegible, suffer water damage, fire damage, etc. As long as you have a sensible backup and conversion strategy, etexts never need to decay (of course, if you do not, any individual instance is very fragile indeed) and are effectively immortal.
• availability: etexts are available wherever you may be. I have a library on my cellphone that is (now that I have got rid of many thousands of ptexts that I used to own) considerably larger than my ptext collection that takes up half a room and, as long as I have a network connection, I have rapid access to countless more billions of etexts, including millions of books for (almost) free, hundreds of millions of them if I don't mind paying for them. This is not to mention journals, magazines and plenty of other texts.
• knowledge sharing: DRM aside, it is so much easier to share an etext than a ptext. etexts are vastly superior for spreading knowledge around. They are non-rival goods where giving an etext away does not take it away from you. 

We have only just reached a point where the benefits of ebooks outweigh those of pbooks, but the gap is widening fast and accelerating. ptexts will never go away in my lifetime and may even experience growth in some areas for a little while to come, thanks to centuries of momentum and the deliberate crippling and price-fixing of etexts by avaricious publishers, but we have already reached a turning point and ptexts are now becoming a niche product that is on its last legs for most purposes. I don't think we will lose much in the long run as we get rid of (most) printed paper, but we will gain a huge amount.