You start more than you finish because your open loops are all legitimate work competing for the same hour, and each switch leaves the last job interrupted. The fix is one finish line per block: pick one piece of work, define what done looks like, and park everything else with a written plan.
It's 4pm on a Tuesday. The other client's deck sits in one tab with slide six half-built. The quarterly report stops mid-sentence in a Google Doc, and a PR review waits in a third tab while Slack pings you about it. You worked all day, and you can't name one thing you shipped.
Each of those tabs holds real work, so a blocklist has nothing to block. I call the pattern fragmentation, and I made the long case for it in The Fragmentation Problem.
You switch before the work reaches a stopping point, and the work you left stays open. Gloria Mark, Victor Gonzalez and Justin Harris shadowed 24 information workers for more than 700 hours, noting each action to the second. In their CHI 2005 study, interruptions broke into 57% of the work segments they logged. People spent an average of 11 minutes 4 seconds in one "working sphere", their term for a unit of work like a project or a client, before moving to another.
People did come back to most of that work, by a long route. They picked up 77.2% of interrupted work the same day, but first they worked in an average of 2.26 other spheres, according to the same paper. The other 22.8% sat untouched for the rest of that day.
You may have read that your brain holds on to unfinished tasks better than finished ones. Bluma Zeigarnik reported that memory advantage in 1927, and researchers have struggled to repeat it since. A 2025 meta-analysis by Romain Ghibellini and Beat Meier pooled 37 later publications and found people recalled interrupted and finished tasks at about the same rate: a ratio of 0.99.
The same meta-analysis found support for a quieter effect that Maria Ovsiankina described in 1928: the urge to go back. Across 21 publications, people resumed an interrupted task about 67% of the time when they had the chance, well above the 50% you'd expect by chance (Ghibellini and Meier, 2025).
My reading of that, applied to your Tuesday: each half-built job tugs you back toward it. With one job open, the tug helps you finish. With the deck, the report and the review all open, you feel three tugs at once and spend the afternoon going back to whichever job comes to mind.
Procrastination means delaying a task you intend to do. Fragmentation feels different from the inside: you work all afternoon and keep changing which job gets the hours.
Some of it is avoidance, though. You reach the hard paragraph in the report, and the PR review starts to feel urgent. One test helps: note where each switch lands. If it lands on the easier job, you're avoiding the hard one. If it lands on whatever pinged last, you're fragmented, and most weeks hold some of both.
Underneath both sits a planning error. In Buehler, Griffin and Ross's 1994 study of the planning fallacy, 37 psychology students predicted they would finish their honors theses in 33.9 days. They took 55.5 days on average, and 29.7% finished by their own best estimate. Even their worst-case forecasts averaged 48.6 days, still short.
You plan each job at its best-case length, so you start more jobs than the week can hold and carry the overflow into next week. Buehler's team also tested a fix. In the paper's fourth study, students who linked a computer assignment to their past experience with similar work forecast it with no optimistic bias, and 60.0% finished within their forecast, against 29.3% of a control group. They didn't finish any sooner. They planned for the time the work took.
The workers in the CHI 2005 study juggled an average of 11.7 separate working spheres each. In that sample, a dozen live projects was ordinary. You can't cut your list to one, but you can change what the other eleven cost you while you work on the twelfth.
E. J. Masicampo and Roy Baumeister measured that cost in 2011. In the first study of their paper, 73 students wrote about two important tasks they still had to do, then read a passage. Among students who made no plan, 65.2% reported their minds wandering during the reading. Among students who wrote down when, where and how they would do each task, 33.3% did.
The planners also reported no more intrusive thoughts about their tasks than a control group who had written about tasks already done. Those students hadn't done the work. They had decided when they would, and in this study that was enough to stop the intrusions. It's a set of lab studies with undergraduates, so treat it as a strong hint about your own desk.
Use one finish line per block. Six steps get you there:
List each open loop. Write down all the live work on one page: the deck, the report, the PR review, the Slack thread you promised to answer. Seeing it in one place shows you the real size of the pile.
Pick one and write its finish line. A finish line is an output someone else could check, like "section 3 of the report sent to Dana". "Work on the report" fails that test.
Give each other loop a when. Next to the deck, write "Thursday 10am, after the client call". Next to the PR review, write "today, 3pm". That's the plan-making step from Masicampo and Baumeister's paper. You still owe the work, and now you've decided when you'll do it.
Size the block from your last real one. Recall how long the last similar section took and book that much time. Your best-case guess is the one the thesis students missed by about three weeks, and linking the forecast to past work is the step that removed the bias in Buehler's fourth study.
Write an if-then for your usual exit. "If I want to check the PR mid-paragraph, I'll add it to the list and finish the paragraph." Peter Gollwitzer and Paschal Sheeran pooled 94 independent tests of plans in this if-then form and found a medium-to-large effect on reaching goals, d = .65 (their 2006 meta-analysis).
Stop at the line, then switch on purpose. Close that tab, tick the item off and pick the next loop from your list. If you have to stop short, write one sentence on where you left off. Mark's team noted that a blinking cursor at the end of your last word helps you find your place, and a screen full of new windows hides it (CHI 2005).
If the deck pulls at you mid-block, look at your list. You already gave it a time.
It can. The CDC lists finding it hard "to organize or finish a task" among the signs of inattentive ADHD. If you're an ADHD worker, the method still applies with smaller pieces: a finish line you can reach before lunch, and the parking list on paper beside your keyboard. An app can't assess you, and a clinician can. There's a separate guide to ADHD focus tools for Mac if you want the software side.
I built Hugo, an AI focus app for Mac, around one goal per session. You start a session by giving it the finish line from step 2. You can type it, or pick it from your own list in Calendar or Reminders on your Mac, Todoist, TickTick or Notion. Hugo reads those lists; it never changes them.
In Hugo AI mode, the AI judges each new tab, AI chat and app against that goal. Say you're on the report and your hand opens the other client's deck. Hugo AI weighs that tab against the report, and if it can't tell, it asks how the page helps your goal. You type a real reason. A reason that serves the report gets you in. "The deck is due Thursday" is true, and it doesn't serve the report, so it doesn't get you in. "Open anyway" sits at the last step, so the final call stays yours.
On the default Standard intensity, ending a session before its time means writing a short signed reflection first. That's step 6 with a witness: if you leave the report short of its line, you say why, in writing, before you go.
Hugo runs on Mac. The 14-day trial covers everything, needs an email and asks for no card.
Pick the finish line. Hugo keeps you on it. Download Hugo.
Ready to stop losing time to distractions?
Hugo uses AI to keep you on task. No blocklist required. Set your goal, and Hugo evaluates every tab in real time.
Set it once. Hugo keeps you on it. 14-day free trial, no card.