Top 10
SuperMemo 16 is faster and more reliable than SuperMemo 15. It works better in Windows 8.1 and with Internet Explorer 11. It conglomerates all incremental learning techniques. It requires fewer clicks to access individual options. It adds dozens of options for improving incremental reading, incremental video, incremental audio, and visual learning. In the long list of changes, some minor unsexy improvements tend to make a world of difference. Here is our vote for improvements that are most likely to make a big difference in your learning:
- Incremental video on local files
- Supplementing the interface by the mouse-oriented learning bar
- Fast search & replace
- Add all to outstanding along with a dedicated icon in the browser
- Global paste in visual learning
- Deleting processed texts with a single keystroke in the Commander
- Speed
- More reliable use of Internet Explorer for handing HTML pages
- Sorting elements by text size in the browser
- Fast YouTube imports
New functions
Incremental audio/video
Incremental video and incremental audio largely complete the set of media available in incremental learning, incremental processing and incremental writing. This leaves the user of SuperMemo with no excuses for using paper, DVDs, cassette tapes, mail files, or other media in keeping archives. All knowledge, information and valuable data can now be stored in SuperMemo and processed incrementally (from knowledge, through picture albums, to family archives, to home video, to university lectures, and more).
- Incremental video with local files. If you do not want to rely on YouTube, or have lectures that cannot be uploaded, or want to process your own videos, you can use MP4 now for incremental video (some other formats are also supported, e.g. MOV, WMA, MPG, AVI, etc.). Not all file formats are supported, however, most formats can easily be converted to MP4 (e.g. with Recode). Before you commit to SuperMemo and a specific format, be sure to give it a test with a trial version
- New Extractor toolbar is available amongst media panel options in sound and video components. Note that this toolbar makes the old option Extract selection obsolete. As a result, that option has been removed from SuperMemo. The Extractor toolbar was modelled after the incremental video panel for processing YouTube videos
- You can define Video Extract template to have your template used for video extracts. If this template is absent, default video template will be used with the background color changed to yellow. The same holds for YouTube extract except for the template name that is YouTube Extract. In the simplest case, redefine your Video or YouTube template after saving it under the new name
- Full screen option for video files. This option adjusts the video size to the size of the video component (i.e. not the size of the screen). The actual full screen is also possible (e.g. with View file)
- YouTube videos get embedded in documents exported as HTML
- The Extractor toolbar includes buttons for rewinding one frame back or one frame forward. Unfortunately, MCI commands do not always execute that command reliably and you may skip a few frames in the process
Speed
- Much faster HTML handling in incremental reading (importing large articles, extracting from large articles, filtering, Wikipedia formatting, etc.)
- Speeding up procedures for converting relative and absolute links in HTML (e.g. loading image rich texts may be up to 35x faster)
- Speeding up tag processing with low-level optimized procedures that are now up to 2000x faster. As a result, long Wikipedia articles import, load and split up to 10x faster
- Faster search and replace procedures on long texts. As a result, loading long plain texts that need to be converted to HTML load twice faster
- No style or paragraph filtering for Wikipedia articles speeds up processing without compromising editability (with new event sinking)
- Removing spacing and styles from filter defaults. Long and richly formatted articles import up to 24x faster
- Old text formatting styles are no longer converted to new styles resulting in a speed gain
- Faster title processing (very long text would slow down all operations where shortened title was needed)
- Matching words is up to 10x faster on long texts
- Faster procedures for counting and sizing files and directories
Search and replace
- Search & Replace for the entire collection now runs in assembly language (shortcut Ctrl+R). Global search and replace is now over 100x faster on large collections with sparse replaces. Have you have ever wondered why Search could take a second to rush through your entire gigabyte collection, while Search & Replace was so slow? Why did you need to resort to running search and replace through search results in the text registry? The new assembly language procedure is as fast as the search procedure. Results of the replacement can be inspected and/or reversed using the first text registry subset
- Search & Replace can now be used in a browser subset (e.g. if you misspell a sentence and generate a couple of cloze deletions that propagate the error)
- Search & Replace can now be used in all registries (e.g. to change names of images, etc.). Results can be reversed with the use of the first registry subset
- Search & Replace now also works on recursive replacements (i.e. you will not see Cannot run Replace All on infinitely recursive replacements). This is done by doing two-step replacement with an intermediary and storing the processed entries in the done queue
Incremental reading
- Delete before cursor command makes it easy to work with badly formatted texts. This sounds like an unexciting function. However, if you use incremental reading, you will soon wonder how you ever lived without it! This function works on raw HTML and is independent of the vagaries of DOM and Internet Explorer. This means that you won't be foiled by bugs in the Explorer or quirks in HTML. Once you locate the information you need, you can get rid of the preceding garbage with a single keystroke
- Delete after cursor deletes all texts from the cursor to the bottom of the article (without affecting references)
- custom strings can be used in article splits
- adding blue headlines to split articles
- topics and items can be sorted using their own degree of randomization
- removing leading HTML tags from texts in places where they could not be reached with a selection and Remove formatting (e.g. Remove Formatting : Front in the Commander, or Text : Convert : Clean lead HTML on the HTML component menu
- <SPAN class=cloze> is used to mark cloze deletions for cleaner and customizable items
Wikipedia
- Wikipedia thumb pictures can now be downloaded in full resolution in image downloads
- Wikipedia redirects are stored as the proper URL to prevent duplicate imports (the redirect is stored as a comment for further reference)
- equivalent, but differently encoded URLs are reported as duplicates. For example, http://en.wikipedia.org/wiki/Projection_%28linear_algebra%29 will be detected as a duplicate of http://en.wikipedia.org/wiki/Projection_(linear_algebra)
- Wikipedia MBoxes are formatted out with an ignore style for easy removal from the collection
- empty subsections do not create new elements at split
- show and hide navigation toggles are filtered out
- updated filter removes Edit links
- updated filter removes v*t*e templates
- jump to: navigation, search in Wikipedia imports would not filter out due to adding an extra argument to the DIV tag by Wikipedia
- importing also via https:// protocol
References
Visual learning
- Pasting pictures can now be local or global. In a global paste, the contents of the clipboard replaces the image in the registry (not just the image in the component). This means that all elements that use the image will now use the newly pasted version.
- Pasting pictures reuses the first existing image component first or squeezes in an extra image component. Successive pastes tile the existing images in their originally covered area of the element window
- Paste over in the image registry allows of replacing pictures with a new version taken from the clipboard
- Pictures can be pasted to image registry as new objects (instead of being pasted to elements or components)
- Ctrl+V and Shift+Ins in the image registry allow of inserting new members or overwriting old members using pictures stored in the clipboard
- Picture import mode in addition to YouTube and Wikipedia modes in web imports.
- Compress images in the Commander can be used to save space and to convert poorly showing GIFs or PNGs to straight JPGs. It can also remedy some CRC problems in PNG files sent via MIME in EML imports (all files are converted to JPG at compression)
- Page of images imports reuse the first existing image area for imports instead of relocating other components
- Download images displays the total count of available images in a given HTML page
- Image imports support Google Images without the need to click View image in Google
- Fit to picture allows of economizing more element area space when fitting multiple pictures
- Support for TIFF format
- GPS data in JPG images can be used to point to the picture taking location in Google Maps
Interface
The new SuperMemo is not more beautiful and may even appear cluttered. However, the new cluttered looks are compensated well enough by a dramatic increase in usability, esp. for those who prefer to use a mouse, or are new to the program. In addition, even seasoned pros will appreciate shorter click-paths as most options are now available with just 2 clicks. Mouse-oriented learning bar makes it easier to use SuperMemo with a single hand.
- navigation bar for finding things in the collection, and learning bar to read, learn and edit the collection
- all toolbars are now also available from buttons on the above two
- most important options are listed in popup menus available from individual buttons on the toolbars
- learning bar makes it easier to access most function of the program from simple menus
- icon set and toolbars redesigned for better usability
- unlike MS ribbons, you always have the most often used buttons in plain view, unless, on a less frequent occasion, you click buttons that lead to other toolbars
- toolbar dock and floating toolbars are no longer needed, however, they have been retained for old timers who do not like to change their habits. All traditional menu structure is retained for the most advanced functions (main menu, element menu, component menu, etc.)
- Keep docked keeps toolbars docked. At the beginner level, the default setting for toolbars is Keep docked. In this setting the toolbars do not float and cannot be dragged away from the dock. This is probably how SuperMemo will work in the future
- adding new components from the Compose toolbar has been simplified. One click is enough instead of the previous 3-click scheme. The added component is set in the resize mode for convenient placing in the element area. When successive elements of the same time are added, they are tiled in the set area
- each collection can use its own background picture that helps distinguish between collection in thumbnails used by Windows 8 (e.g. when switching between applications)
- alarms has been integrated with the Alarm toolbar instead of the disruptive dialog box
- increased range of allowed interface font sizes
- when using registries or the browser, the toolbars are not swept to the side, and are docked instead
- closing the toolbars saves their status in the current layout (without the need to save the layout)
- Contents and Browser layouts do not save or restore their own toolbar layout
- when a toolbar is first display, it is in the docked state (rather than a floating state). If the dock is not available, user can chose if it should be shown
Commander
New commands:
- Add pronunciation
- Ancestors: Show
- Compress images
- Delete after cursor
- Delete before cursor (one of the most useful new functions in SuperMemo 16!)
- Done!
- Duplicate
- Highlight
- Remove formatting: leading HTML
- Remove formatting: plain text
- Remove formatting: selected text
- Send E-mail
- Sort repetitions
- Tile components
- Unzoom image
- Zoom image mode
Browser
- Browser supports sorting elements by text size: e.g. if you get lost in processing very long articles or lists and you need to begin with the least processed fragments
- elements can be sorted by the date of the first repetition (or the date on which the element had been introduced into the collection)
- Child : Extracts can be used to work with extracts as opposed to full articles
- Child : Articles can be used to work with full articles as opposed to article extracts
- repeating the sort by contents or size results in a reverse sort (as used to be the case with other sorting criteria)
Templates
Registries
- inserts of objects from the registry tile new components (instead of rearranging the existing components)
- status bar displays text sizes for texts stored directly in the registry (not only those in HTML files, etc.)
YouTube
- YouTube references download in a separate thread while watching videos. If there was a poor bandwidth or the YouTube server was slow, the import of just a few dozen videos might take several minutes, while now it will take just a few seconds. If references cannot be downloaded, the element will not be stamped with the unfriendly "reference download error". Instead, SuperMemo will attempt to connect with the server next time you view the video. Moreover, you can re-download or update all references with Reference : Download on the HTML component menu
Statistics
- Protection in statistics shows the degree of processing the top priority material on a given day. It shows the priority of the highest priority item and the highest priority topic still in the outstanding queue
- Topic and item statistics have been separated for all entries where averaging the two provides misleading values (e.g. average repetitions count, interval, last repetition date, and more)
- the Outstanding parameters displays additionally the total size of the outstanding queue if it includes elements that were included with Add to outstanding. The same change shows on the status bar
- A-Factor and Repetitions in statistics are displayed with I= and T= prefixes for items and topics to avoid confusing the two groups of elements
- Statistics window font and row height adapts to screen resolution and the size of the statistics window
- new browser statistics: total value of tasks, total cost of tasks, average value, average cost, average yield (value/cost), etc.
- text size statistics are added by default in browser statistics
- Analysis : Use statistics use meaningful names (instead of just file names)
- sleep statistics are updated instantly when deleting or adding sleep blocks
Incremental mail processing
- Mail import from Windows Mail and Windows Live Mail as well as other mailers that support EML
- MIME imports: supported HTML, plain text, line-folding, utf-8, u-16, ISO-8859 (1 to 15), especials, base64, printed-quotable, multi-part messages, jpeg, gif, png, bmp, tiff, multiple audio formats (incl. mp3, wma, wav, etc.), multiple video formats (incl. mp4, wmv, etc.), binary attachments, etc.
- EML text format (files are stored in the original format and displayed as HTML)
- EML to HTML conversion (automatic at mail editing)
- imported mail may reside in the original folder as duplicate imports are prevented (auto-archiving is not needed for EML imports)
- formatted EML header is added to mail displayed as HTML
- mail import report (errors, mail skipped, mail imported, etc.)
- detecting default folders for Windows Mail and Windows Live Mail (only in Windows Vista and later)
- mail recipient name added to comments in mail references
SleepChart
- dense view in the sleep log chart makes it possible to see annual trends
- Circadian graph can be used to inspect circadian rhythms of learning (in the same way as in the analysis of sleep cycles)
- Sleep length vs. nap length graph makes it possible to compute nap compensation ratio in sleep deprivation
- Periodogram: 32768 samples at 3 min. intervals give optimum display for 500 points in the graph (113 days). Exact period of peaks can be shown in the graph
- it is possible to inspect sleep data of other users of SleepChart (in addition to one's own data). Naturally, all learning estimates are only possible if the relevant collection is inspected as well
- green 24h balance line in the circadian graph makes it easier to optimize the length of the waking day in sleep phase disorders
- Fourier transform export from SleepChart to Excel
- sleep vs prior sleep graph is available on the SleepChart : Learning tab with a click (undocumented)
- middle-of-the-night provides for better wake and sleep estimations over moving averages
- homeostatic recall data for Excel now include the full name of elements taking part in repetitions that contribute data
- better heuristic for differentiating between circadian zones of orphaned blocks in fragmented sleep (lesser impact of block length and wake time)
- REM sleep heuristic
- the mousewheel can be used to scroll the log
- SleepChart allows of a dedicated folder for Excel imports and exports
- converting SleepChart to use TDateTime data type internally (as opposed to low precision floating point). The data format on the disk has not changed. SleepChart converts 18-byte records back to the old 12-byte format (i.e. different versions can exchange sleep logs)
Plan
- Changing the starting hour of the schedule allows of four modes of the fixed-point update: (1) unfix, (2) shift equally, (3) keep in place, or (4) shift in proportion (to make sure time anchors keep the schedule in the desired frame). On different days, you may need a different mode. All these mode have their place in optimizing your day
- Adjust uses the ActLen column instead of the OptLen column in making the length adjustments. This makes it possible to account for fixed-time activities and make sure that all slots have their efficiency (% column) set at 100% at the beginning of the execution of the schedule
- % column in Plan compares actual length with the planned length, not with the optimum length (the two should be the same in the adjusted schedule)
- drop in execution efficiency can be registered after executing each slot that lasted longer than planned
- Split uses the actual activity length rather than the planned length. This provides for more realistic time allocations for compressed slots even though the split itself will disadvantage the split slot in reference to the remaining slots
- Split allows of splitting zero-length activities (e.g. after adding a new slot and before determining its optimum length)
- Display daily totals is on by default for Totals
Reliability
- New way of sinking events in HTML documents should make SuperMemo infinitely more reliable in processing HTML (problems with mouse click errors, mouse wheel, mouse responsiveness in tables, etc.)
- recompiled with Delphi XE3 (SuperMemo 15 was written in Delphi 2010)
- optimized for Windows 8 Desktop (SuperMemo 15 would run best in Windows Vista). As of November 2013, it should work best in Windows 8.1 with Internet Explorer 11. See: Minimum requirements
- most frequently encountered errors allow of an instant Google search for solutions listed in FAQs or in SuperMemoPedia
- better filespace error diagnostics
- mail imports generate a file with the list of encountered problems (e.g. duplicate imports, file locks, corrupted attachments, etc.)
- better resistance of visual browser to script errors and other problems that slow the loading of elements
- better resistance to errors in media files (e.g. sound, video, etc.), esp. floating-point errors generated by video drivers
- PNGs that import with data errors (e.g. from EML mail) are converted to BMP
- no error report when temporary files cannot be deleted (they are cleaned up at next Repair collection)
- dedicated procedure for handling floating point checks and errors (e.g. as generated by video drivers)
- at opening the database, SuperMemo displays a warning if the setting of the secondary storage is invalid
- Glossary:Filespace recovery is blocked if the secondary storage is set to an invalid folder. Options dialog opens on the Access page to assist in choosing the correct setting
Annoyances
- In SuperMemo 15, HTML components would not allow of empty texts and would generate random strings to newly added empty items. This was very confusing for beginners and very annoying for the pros
- "Processing tags" is displayed on the status bar instead of an annoyingly blinking dialog box
- cancelled extracts would not be marked with ignore style
- information message Reusing an existing registry member was confusingly displayed with a yellow background like an error
- A-Factors could be displayed with localized formatting (e.g. as AF=3,1 instead of AF=3.1)
- reference tags could not start nor end with quotations marks
- the annoying Internet Explorer error mshtml.dll refuses to paste is not displayed (it is ignored instead with no negative side effects)
- if you hate the behavior of Enter over read-points and would prefer to delete the selection rather than to move on in the repetition cycle, you can hack it out by changing a setting in supermemo.ini files (set Allow Read-Point Enter to zero)
- Warrior layout no longer plays with toolbars and keeps them docked
- many shortcuts would not work in the browser when the menu was docked in the element window
- many shortcuts would not work in the contents window when the menu was docked in the element window
- Del key would not work when editing Web Import title
- double confirmation Clear all fields?, and Save Changes? is not really needed in Plan : New
- if a collection cannot be opened due to a wrong version of SuperMemo, it is deleted from the pick list so that another one could be tried
- File browser does not ask Delete KNO? after getting No to Delete collection? (it is unlikely you will want to delete the KNO file without deleting the collection)
- Template registry now does not ask about template impose or detach if the template is not used by any elements
- Edit : Add components : Other opens the compose toolbar on the learnbar (not a floating toolbar)
Bugs
The list of bug fixes is the longest in SuperMemo history (over 1,000 positions). All that despite the fact that SuperMemo 15 was probably the best debugged version before SuperMemo 16. Here is only the list of most painful or annoying bugs:
- mouse wheel would be erratic when working with HTML texts
- click on a scrolled HTML page would often misplace the cursor
- some URL would be wrongly marked as import duplicates
- some Unicode text would not be found with Match Word in Find elements (e.g. "attaché")
- search would not find Borbély’s if the apostrophe was represented as #8217 (HTML: ’)(’)
- auto-sort would not update statistics
- mail import folder would be created even if no mail was chosen for import
- HTML tags would affect similarity ratings in XML text updates
- circadian graph would show a Range Check error if the last day was greater than the size of data
- inserting sound from the registry would not stop the registry player and cause file access conflicts
- F11 for a random jump would not be blocked when using Plan
- in some circumstances, neighboring graph calibrations would show the same value with too few decimal points in formatting
- Workload Graph would not extend into the past for past Repetitions or Retention modes
- Workload export to Excel would not use double quotes to enclose the dates
- toolbar dock query would return false if the main dock was invisible. This would save toolbars docked on the invisible main window in the state registered as floating
- trailing wake averages where computed with a substantial margin of error in SleepChart
- manual changes to the length of the Plan schedule could result in a range check error
- background change status set to "choose" might have no effect (i.e. background not saved)
- changing Start in the Plan schedule would ask for a shift in fixed times even if not fixed times were defined
- incomplete SleepChart data would produce errors in circadian alertness estimations
- Alt+Left might result in Access denied error when traversing sites on the web
- Alt+Left on the external site might result in the use of the wrongly cached web page
- it was possible to save an external web page over the original source
- mail address would not be retrieved if the author's name was followed by a new line
- missing SuperMemo background picture error would never self-correct. Now it displays just once and clears the background
Warnings
- note that global replace of images in the registry is pretty powerful and thus dangerous. If you have become accustomed to local changes, you may mistakenly choose a wrong option (global instead of local, just one radio button below). Despite double confirmation, you can fail to pay attention and replace an image used in many places in your collection. You can undo that option only by pasting back the original image (e.g. from a backup).
New Limitations
Please read these carefully if you plan to upgrade from older versions of SuperMemos!!! Not all old functions are supported in SuperMemo 16:
- US English locale is enforced throughout the program to avoid multiple past bugs (e.g. A-Factors formatted as 1,34 would not transfer between platforms, or file using month names would not be recognized when changing the locale, e.g. when switching computers, etc.)
- To speed up SuperMemo, old text formatting styles are no longer converted to new styles. If you upgrade a very old collection and you want your styles converted, you might use the collection first with SuperMemo 15 to get your styles fixed automatically during edits. This is mostly an esthetic issue. If you upgrade directly to SuperMemo 16, your old and new texts might be formatted differently (e.g. different styles for clozes, extracts, etc.).
- Incremental video requires computers with significant processing power. It may be annoyingly slow on older or weaker PCs. This comes from the size of video files, not from limitations of SuperMemo itself.
- HTML components have been redesigned radically to improve their reliability. However, this entailed adding a limitation: HTML components cannot be resized in editing mode. To resize, you need to switch to drag and resize mode (e.g. with Ctrl+E or Alt+click). In the long run, you may like this change as this prevents accidental resizing your template when hitting the border accidentally in incremental reading
- cursor in HTML remains in the old position when switching to the display mode and back to editing - this can initially cause typing errors if you have got accustomed to the old mode (cursor set at the beginning on each entry to editing; or at the reading point if there is one)
- no automatic upgrade of RTF texts to HTML. RTFs are displayed in HTML marked with a red border, but retain their original formatting and can be converted to HTML upon first edit
- some options may not work in older Windows. For example, in Windows XP: no visual browser, no default mail folder detection. In Windows Vista: no checked menu items, no transparent toolbars
- search and replace in references registry, works on members, not reference texts
- search and replace does not affect EML files (they are always identical to the original mail until they are converted to HTML, e.g. with the first edit, or Ctrl+Shift+F12)
- Find in Search and Replace is retained but is of the same old speed. It searches the collection along physical numbers, not along the knowledge tree.
- Semi-Copy option has been removed. With an increasing reliance on HTML, the \elements folder is in wider use than in the old times when it was mostly used for multimedia files stored on CD-ROM
- due to the extent of MIME specifications, some mail might import incorrectly (esp. with less popular charsets or file formats). You can always open the original mail with F9. Please remember to forward for debugging if possible!
- extracting texts to comment registry in collection filtering is no longer supported. This option was rarely used and would slow down filtering
- Transcription character map has been removed. Phonetic transcription is now Unicode-only. We recommend Type-It from Antimoon.com instead